• Defeat Python "Virtual Environment" in Fedora ?

    From c186282@[email protected] to comp.os.linux.misc on Sun Mar 8 05:31:16 2026
    From Newsgroup: comp.os.linux.misc

    Hmmmmm ....

    F43

    Trying to do basic 'system' commands - like 'reboot'
    or 'kill' and even running root shell scripts as
    root .....

    Alas Python is no longer "real" ... lives in a
    horrible "virtual environment" that seems to lock
    out a lot of useful stuff "for your safety".

    There SHOULD be a CL option "python3 -NVE" or
    whatever to totally disable the VE. There isn't.

    Been searching ... NO fixes that actually WORK
    worth a damn. "disable" - is that something you
    have to do in a CL evocation ? No such command
    in the real Linux system. Sorry, I write Python
    scripts that are NOT ultra-transient - saved pgms.

    Do I have to go to Pascal/'C' again just to Get
    Stuff Done ??? I will if necessary.

    What a WASTE of an otherwise good language ... what
    kind of MORONS did the org hire ???????

    Did try "PyPy" ... but there are a number of lib
    differences - first bitch was that there's no
    ".isnumeric" thing. But, is PyPy the way to go now ?
    I can write an "isnumeric" easily enough ...

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Sun Mar 8 18:57:07 2026
    From Newsgroup: comp.os.linux.misc

    On Sun, 8 Mar 2026 05:31:16 -0400, c186282 wrote:

    Alas Python is no longer "real" ... lives in a horrible "virtual
    environment" that seems to lock out a lot of useful stuff "for your
    safety".

    I started using venvs long ago but it is getting more painful. One thing
    I've started doing is

    python3 -m venv foo --system-site-packages

    At least on Linux it's symlinks and not copying all the files. That saves
    a lot of annoyances. After activating

    pip freeze -l > requirements.txt

    You should do that anytime you install something with pip. The '-l' will
    only list what you've installed in that environment. Without it lists all
    the system packages too.

    I've been burned when a distro upgrade introduces a new Python version or
    if I want to move a venv. After creating a new venv and activating

    pip install -r requirements.txt

    pulls in everything you'd installed in the original. Then you can blow it away.

    Sometimes 'sudo apt install python3-some_package' works and adds it to the system packages but it doesn't work for a lot of them.

    There are some other flags that are useful like 'pip install foo --
    upgrade'

    I've also been playing with uv.

    https://docs.astral.sh/uv/

    It makes it easier to work with multiple Python versions, like nvm for multiple node.js versions. I haven't tried it on the Fedora box but the upgrade to 43 installed Python 3.14. Fine and dandy except the maintainers
    of PySide6 and some other packages hadn't upgraded to work with 3.14. It
    took me a while to figure out the problem from the errors thrown by pip install.

    All in all the environments are a good thing. I think virtualenv will
    still work although venv is preferred.

    Personally I stay away from conda. There are packages that aren't
    available from conda-forge and it doesn't play very well with a normal
    Python install.






    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc on Sun Mar 8 19:54:49 2026
    From Newsgroup: comp.os.linux.misc

    rbowman <[email protected]> writes:
    On Sun, 8 Mar 2026 05:31:16 -0400, c186282 wrote:
    Alas Python is no longer "real" ... lives in a horrible "virtual
    environment" that seems to lock out a lot of useful stuff "for your
    safety".

    I started using venvs long ago but it is getting more painful. One thing I've started doing is

    If c186282 actually means venvs, and doesn’t want to be in one, then ‘deactivate’ is all they need.

    However, the idea that Fedora 43 pushes Python into a venv by default
    does not pass the sniff test (and doesn’t seem to be true in the
    container image, which has what looks like a perfectly normal Python
    3.14).

    Perhaps they have confused it with something else.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Sun Mar 8 21:44:40 2026
    From Newsgroup: comp.os.linux.misc

    On Sun, 08 Mar 2026 19:54:49 +0000, Richard Kettlewell wrote:

    rbowman <[email protected]> writes:
    On Sun, 8 Mar 2026 05:31:16 -0400, c186282 wrote:
    Alas Python is no longer "real" ... lives in a horrible "virtual
    environment" that seems to lock out a lot of useful stuff "for your
    safety".

    I started using venvs long ago but it is getting more painful. One
    thing I've started doing is

    If c186282 actually means venvs, and doesn’t want to be in one, then ‘deactivate’ is all they need.

    However, the idea that Fedora 43 pushes Python into a venv by default
    does not pass the sniff test (and doesn’t seem to be true in the
    container image, which has what looks like a perfectly normal Python
    3.14).

    Perhaps they have confused it with something else.

    'pip3 install ruff' on Fedora gets 'defaulting to user installation
    because normal site-packages is not writable'.

    Ubuntu is more verbose.

    $pip3 install ruff
    error: externally-managed-environment

    × This environment is externally managed
    ╰─> To install Python packages system-wide, try apt install
    python3-xyz, where xyz is the package you are trying to
    install.

    If you wish to install a non-Debian-packaged Python package,
    create a virtual environment using python3 -m venv path/to/venv.
    Then use path/to/venv/bin/python and path/to/venv/bin/pip. Make
    sure you have python3-full installed.

    If you wish to install a non-Debian packaged Python application,
    it may be easiest to use pipx install xyz, which will manage a
    virtual environment for you. Make sure you have pipx installed.

    See /usr/share/doc/python3.13/README.venv for more information.

    note: If you believe this is a mistake, please contact your Python installation or OS distribution provider. You can override this, at the
    risk of breaking your Python installation or OS, by passing --break- system-packages.


    $ sudo apt install python3-ruff
    [sudo: authenticate] Password:
    Package python3-ruff is not available, but is referred to by another
    package.
    This may mean that the package is missing, has been obsoleted, or
    is only available from another source

    Error: Package 'python3-ruff' has no installation candidate


    https://peps.python.org/pep-0668/

    https://packaging.python.org/en/latest/specifications/externally-managed- environments/#externally-managed-environments

    says it clearer:

    "This specification defines an EXTERNALLY-MANAGED marker file that allows
    a Python installation to indicate to Python-specific tools such as pip
    that they neither install nor remove packages into the interpreter’s
    default installation environment, and should instead guide the end user towards using Python Virtual Environments."


    You don't necessarily have to use a venv if what you're doing is covered
    by the standard Python packages but there is a strong attempt to herd you
    into a venv.




    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc on Sun Mar 8 22:05:37 2026
    From Newsgroup: comp.os.linux.misc

    rbowman <[email protected]> writes:
    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to override the restriction if appropriate.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Charlie Gibbs@[email protected] to comp.os.linux.misc on Mon Mar 9 00:27:02 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-08, Richard Kettlewell <[email protected]d> wrote:

    rbowman <[email protected]> writes:

    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling packaging systems’ mess, and looks like it’s straightforward to override the restriction if appropriate.

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but
    this discussion is scaring me off. To quote the old Depression-era
    saying: "Is this trip really necessary?"
    --
    /~\ Charlie Gibbs | Growth for the sake of
    \ / <[email protected]d> | growth is the ideology
    X I'm really at ac.dekanfrus | of the cancer cell.
    / \ if you read it the right way. | -- Edward Abbey
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Sun Mar 8 21:14:00 2026
    From Newsgroup: comp.os.linux.misc

    On 3/8/26 20:27, Charlie Gibbs wrote:
    On 2026-03-08, Richard Kettlewell <[email protected]d> wrote:

    rbowman <[email protected]> writes:

    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to override >> the restriction if appropriate.

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but
    this discussion is scaring me off. To quote the old Depression-era
    saying: "Is this trip really necessary?"

    At some point the 'idealists' can degrade and ruin
    a perfectly good thing. The idea of a Python VE is
    a stone on this road to ruin.

    Oh well, I've done a lot of Pascal and 'C' ... so
    as needed I can go there. Python IS super handy
    however and has a a lib for everything.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 01:34:35 2026
    From Newsgroup: comp.os.linux.misc

    On Sun, 08 Mar 2026 22:05:37 +0000, Richard Kettlewell wrote:

    rbowman <[email protected]> writes:
    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling packaging systems’ mess, and looks like it’s straightforward to override the restriction if appropriate.

    I've gotten used to it and don't find it cumbersome but I also use dotnet, express, arduino_cli that generate the skeleton of a project.

    I've never tried a package that Fedora wants to install to the user's
    space. Arch generates an error similar to the Ubuntu one. ruff isn't a
    good test on either of those since both distros have a python-ruff package that can be installed with dnf or pacman. I don't know why Ubuntu doesn't since ruff is a general purpose tool like black. The PEP does say distros
    roll their own. As the distros use more Python behind the scenes they have
    a real interest in people not randomly installing to the system packages.

    I fired up the Leap 16 VM and it gives the same external message as
    Fedora. 'sudo zypper install python3-ruff' works on that too.

    There may still be more forgiving distros but with the Debian, Arch,
    Fedora, and SUSE families working mostly the same I think C12345 better embrace venv.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Lawrence =?iso-8859-13?q?D=FFOliveiro?=@[email protected] to comp.os.linux.misc on Mon Mar 9 01:36:55 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 09 Mar 2026 00:27:02 GMT, Charlie Gibbs wrote:

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but
    this discussion is scaring me off. To quote the old Depression-era
    saying: "Is this trip really necessary?"

    I think these people are using lots of custom packages not available
    in the standard distro repos. I haven’t found much need for that
    (except for Jupyter, which is a bit of a monster, so I install it into
    its own virtualenv). In the code I write for myself or for a customer,
    I try to keep external nonstandard dependences to a minimum.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Sun Mar 8 22:11:42 2026
    From Newsgroup: comp.os.linux.misc

    On 3/8/26 21:34, rbowman wrote:
    On Sun, 08 Mar 2026 22:05:37 +0000, Richard Kettlewell wrote:

    rbowman <[email protected]> writes:
    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to override >> the restriction if appropriate.

    I've gotten used to it and don't find it cumbersome but I also use dotnet, express, arduino_cli that generate the skeleton of a project.

    I've never tried a package that Fedora wants to install to the user's
    space. Arch generates an error similar to the Ubuntu one. ruff isn't a
    good test on either of those since both distros have a python-ruff package that can be installed with dnf or pacman. I don't know why Ubuntu doesn't since ruff is a general purpose tool like black. The PEP does say distros roll their own. As the distros use more Python behind the scenes they have
    a real interest in people not randomly installing to the system packages.

    I fired up the Leap 16 VM and it gives the same external message as
    Fedora. 'sudo zypper install python3-ruff' works on that too.

    There may still be more forgiving distros but with the Debian, Arch,
    Fedora, and SUSE families working mostly the same I think C12345 better embrace venv.


    Oddly, in the huge net universe, I can't seem to
    find a good clean generic way to DEFEAT the venv,
    not even for Linux. Most advice mentions CL shit
    that doesn't work/exist.

    Oh, can't "embrace" what THWARTS my needs.

    Back to Wirth and K&R ......

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Charlie Gibbs@[email protected] to comp.os.linux.misc on Mon Mar 9 02:47:51 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-09, c186282 <[email protected]> wrote:

    On 3/8/26 20:27, Charlie Gibbs wrote:

    On 2026-03-08, Richard Kettlewell <[email protected]d> wrote:

    rbowman <[email protected]> writes:

    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to override
    the restriction if appropriate.

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but
    this discussion is scaring me off. To quote the old Depression-era
    saying: "Is this trip really necessary?"

    At some point the 'idealists' can degrade and ruin
    a perfectly good thing. The idea of a Python VE is
    a stone on this road to ruin.

    Oh well, I've done a lot of Pascal and 'C' ... so
    as needed I can go there. Python IS super handy
    however and has a a lib for everything.

    I did write a little Python program that played with the
    various GUI gadgets that it made available (with much less
    effort than doing it in C). It made me think that I could
    write something that my wife might actually be able to use.
    (I was careful to use Python 3, not 2.) It would be nice
    if this remained a viable option; otherwise it's back to
    writing Yet Another C Program.
    --
    /~\ Charlie Gibbs | Growth for the sake of
    \ / <[email protected]d> | growth is the ideology
    X I'm really at ac.dekanfrus | of the cancer cell.
    / \ if you read it the right way. | -- Edward Abbey
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Rich@[email protected] to comp.os.linux.misc on Mon Mar 9 02:59:55 2026
    From Newsgroup: comp.os.linux.misc

    Charlie Gibbs <[email protected]d> wrote:
    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it
    in C). It made me think that I could write something that my wife
    might actually be able to use. (I was careful to use Python 3, not
    2.) It would be nice if this remained a viable option; otherwise it's
    back to writing Yet Another C Program.

    Not necessarially. If the "GUI gadgets" were from the Tkinter package,
    then the GUI gadgets were actually Tk widgets from the TCL world. So
    you could also sidestep all the python venv silliness by writing that
    GUI in TCL instead (and if it's just calling out to other programs to
    do the real work, the amount of TCL necessary to glue a GUI to another
    program is small).
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Lawrence =?iso-8859-13?q?D=FFOliveiro?=@[email protected] to comp.os.linux.misc,comp.lang.python on Mon Mar 9 03:55:54 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it
    in C). It made me think that I could write something that my wife
    might actually be able to use. (I was careful to use Python 3, not
    2.) It would be nice if this remained a viable option; otherwise
    it's back to writing Yet Another C Program.

    C would only be better if you could somehow get by with fewer
    dependencies than with the Python equivalent. I’m not sure how you
    would manage that -- reinvent more of your own wheels, perhaps?
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Mon Mar 9 00:03:18 2026
    From Newsgroup: comp.os.linux.misc

    On 3/8/26 22:47, Charlie Gibbs wrote:
    On 2026-03-09, c186282 <[email protected]> wrote:

    On 3/8/26 20:27, Charlie Gibbs wrote:

    On 2026-03-08, Richard Kettlewell <[email protected]d> wrote:

    rbowman <[email protected]> writes:

    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt >>>>> to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to override
    the restriction if appropriate.

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but
    this discussion is scaring me off. To quote the old Depression-era
    saying: "Is this trip really necessary?"

    At some point the 'idealists' can degrade and ruin
    a perfectly good thing. The idea of a Python VE is
    a stone on this road to ruin.

    Oh well, I've done a lot of Pascal and 'C' ... so
    as needed I can go there. Python IS super handy
    however and has a a lib for everything.

    I did write a little Python program that played with the
    various GUI gadgets that it made available (with much less
    effort than doing it in C). It made me think that I could
    write something that my wife might actually be able to use.
    (I was careful to use Python 3, not 2.) It would be nice
    if this remained a viable option; otherwise it's back to
    writing Yet Another C Program.


    I've written 5000 line Python apps ... one
    smart backup app maybe a bit longer. It is
    a very versatile lang, mostly self-documenting,
    plus tons and tons of online doc/examples.

    But they seem to be ruining it ....

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Mon Mar 9 00:23:47 2026
    From Newsgroup: comp.os.linux.misc

    On 3/8/26 22:59, Rich wrote:
    Charlie Gibbs <[email protected]d> wrote:
    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it
    in C). It made me think that I could write something that my wife
    might actually be able to use. (I was careful to use Python 3, not
    2.) It would be nice if this remained a viable option; otherwise it's
    back to writing Yet Another C Program.

    Not necessarially. If the "GUI gadgets" were from the Tkinter package,
    then the GUI gadgets were actually Tk widgets from the TCL world. So
    you could also sidestep all the python venv silliness by writing that
    GUI in TCL instead (and if it's just calling out to other programs to
    do the real work, the amount of TCL necessary to glue a GUI to another program is small).

    TK is not 'sophisticated', but it WORKS - and, once
    you get a couple of nuances, works very logically.
    A very useful thing is a 'timer/engine' function
    so you can do stuff async without relying on
    immediate user input. I had one app with tons
    of time-dependent functions ... all handled
    by a big "if xCounter>secs" construct. The
    app collected weather, radar, security cams
    and even had a news scroll ... touch to enlarge
    various bits of data. Stuck it in a public
    place in my office ... dunno how many actually
    looked at it but it WAS all there at the
    literal fingertip.

    All ran on a Pi4 stuck to a little monitor.

    The new guy couldn't program six lines of Python
    or much anything else ... he probably trashed
    the thing. Still have the code somewhere.

    DID find that once you create a window it's best
    not to keep deleting/recreating ... just set the
    coords well off-screen and then bring them back
    on-screen when needed. This worked super well for
    that touch-screen app I wrote. Alas, a few TK uses
    do use the horrible evil "lambda" :-)

    QT/GTK stuff ... more complicated. Same stuff, but
    a lot harder to write and with crappier online
    docs. Dunno why some think it's "superior" to TK.
    MAYbe a bit faster, but is that worth the pain ?

    Anyway, for Python I'll stick with TK. For most
    other GUI apps I'll use Lazarus.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc,comp.lang.python on Mon Mar 9 00:28:47 2026
    From Newsgroup: comp.os.linux.misc

    On 3/8/26 23:55, Lawrence D’Oliveiro wrote:
    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it
    in C). It made me think that I could write something that my wife
    might actually be able to use. (I was careful to use Python 3, not
    2.) It would be nice if this remained a viable option; otherwise
    it's back to writing Yet Another C Program.

    C would only be better if you could somehow get by with fewer
    dependencies than with the Python equivalent. I’m not sure how you
    would manage that -- reinvent more of your own wheels, perhaps?

    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Jon Ribbens@[email protected] to comp.os.linux.misc,comp.lang.python on Mon Mar 9 08:16:09 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-09, c186282 <[email protected]> wrote:
    On 3/8/26 23:55, Lawrence D’Oliveiro wrote:
    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:
    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it
    in C). It made me think that I could write something that my wife
    might actually be able to use. (I was careful to use Python 3, not
    2.) It would be nice if this remained a viable option; otherwise
    it's back to writing Yet Another C Program.

    C would only be better if you could somehow get by with fewer
    dependencies than with the Python equivalent. I’m not sure how you
    would manage that -- reinvent more of your own wheels, perhaps?

    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    Python virtual environments do nothing whatsoever to stop you calling
    whatever you like with "os.system()". A Python "virtual environment"
    is simply a separate set of installed third-party library packages,
    to avoid conflicts between the packages used by your program and
    those used by other programs. You seem to be under the false impression
    that it's some sort of chroot/jail/sandbox, and it simply isn't.
    Python has no sandbox feature.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 08:26:05 2026
    From Newsgroup: comp.os.linux.misc

    On Sun, 8 Mar 2026 22:11:42 -0400, c186282 wrote:

    Oddly, in the huge net universe, I can't seem to find a good clean
    generic way to DEFEAT the venv,
    not even for Linux. Most advice mentions CL shit that doesn't
    work/exist.

    I doubt it. I looked at Ubuntu, Fedora, Arch, and SUSE. They differ a bit
    but if you do manage to install something to the system site-packages
    you'll probably regret it. A couple of Ubuntu versions ago 'sudo pip3
    install' would whine but install the package globally.

    According the the PEP the distros have their own Python binaries that may
    be patched from what you get from the python.org download but the distros themselves use Python so uninstalling their stuff and installing from python.org probably would be a disaster.

    Back around 2000 I switched from Red Hat Linux when their release not only broke gcc so it couldn't compile the kernel but their Python broke
    existing scripts.

    I don't know what would happen if you installed 3.13 from python.org on Fedora's 3.14 box and used uv to switch them around. I wouldn't go earlier than 3.13 since I think that's when f-strings were added.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 08:36:04 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 09 Mar 2026 00:27:02 GMT, Charlie Gibbs wrote:

    On 2026-03-08, Richard Kettlewell <[email protected]d> wrote:

    rbowman <[email protected]> writes:

    You don't necessarily have to use a venv if what you're doing is
    covered by the standard Python packages but there is a strong attempt
    to herd you into a venv.

    The design seems better than defaulting to getting into a ‘duelling
    packaging systems’ mess, and looks like it’s straightforward to
    override the restriction if appropriate.

    This thread has been making my eyes glaze over. I was thinking of
    writing a small utility in Python to get to know the language, but this discussion is scaring me off. To quote the old Depression-era saying:
    "Is this trip really necessary?"

    It all depends on what you want to do. A lot of functionality is included
    in the system site-packages or in python3 itself and you can get pretty
    far. For example you could do GUIs with PyTk out of the box afaik. I
    prefer PySide6, the Qt binding, so my venv for that has

    PySide6==6.10.2
    PySide6_Addons==6.10.2
    PySide6_Essentials==6.10.2
    shiboken6==6.10.2

    Start messing around with ML or data analysis and you wind up with

    contourpy==1.3.2
    cycler==0.12.1
    fonttools==4.58.5
    joblib==1.5.1
    kiwisolver==1.4.8
    matplotlib==3.10.3
    numpy==2.3.1
    packaging==25.0
    pandas==2.3.1
    pillow==11.3.0
    pyparsing==3.2.3
    python-dateutil==2.9.0.post0
    pytz==2025.2
    scikit-learn==1.7.0
    scipy==1.16.0
    six==1.17.0
    threadpoolctl==3.6.0
    tzdata==2025.2


    Some of those you don't even install yourself but are pulled in by other packages. pillow is for image handling and I assume at least matplotlib
    needs it.







    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 08:41:58 2026
    From Newsgroup: comp.os.linux.misc

    On Sun, 8 Mar 2026 21:14:00 -0400, c186282 wrote:

    At some point the 'idealists' can degrade and ruin a perfectly good
    thing. The idea of a Python VE is a stone on this road to ruin.

    I disagree. VEs save you from the equivalent of dll hell someplace down
    the road. Dotnet, Angular, Node, and others are similar now. Your project encapsulates the dependencies it needs. Another project may require
    different versions. For example with node anything you install for the
    project goes into node_modules and package.json has the dependencies for
    that project:

    "dependencies": {
    "cookie-parser": "~1.4.4",
    "debug": "~2.6.9",
    "express": "^4.21.2",
    "fs": "^0.0.1-security",
    "leaflet": "^1.9.4",
    "leaflet-gpx": "^2.2.0",
    "morgan": "~1.9.1"
    }


    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 08:52:01 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it in
    C).
    It made me think that I could write something that my wife might
    actually
    be able to use.
    (I was careful to use Python 3, not 2.) It would be nice if this
    remained a viable option; otherwise it's back to writing Yet Another C Program.

    Stick to tkinter and you *should* be good. I say should because 'import tkinter' works on Mint, Ubuntu, and Fedora. On Leap you'd have to do
    'zypper install python-tk' but it would then be global. Arch says it's
    python isn't configured for Tk.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 08:59:16 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 9 Mar 2026 00:03:18 -0400, c186282 wrote:

    I've written 5000 line Python apps ... one smart backup app maybe a
    bit longer. It is a very versatile lang, mostly self-documenting,
    plus tons and tons of online doc/examples.

    But they seem to be ruining it ....

    Welcome to 2026.

    https://www.youtube.com/watch?v=90WD_ats6eE

    The Times They Are A-Changin'

    "Your old road is rapidly agin'
    Please get out of the new one if you can't lend your hand
    For the times, they are a-changin'"
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Nuno Silva@[email protected] to comp.os.linux.misc on Mon Mar 9 09:45:53 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-09, rbowman wrote:

    On Sun, 8 Mar 2026 21:14:00 -0400, c186282 wrote:

    At some point the 'idealists' can degrade and ruin a perfectly good
    thing. The idea of a Python VE is a stone on this road to ruin.

    I disagree. VEs save you from the equivalent of dll hell someplace down
    the road. Dotnet, Angular, Node, and others are similar now. Your project encapsulates the dependencies it needs. Another project may require different versions. For example with node anything you install for the project goes into node_modules and package.json has the dependencies for that project:

    "dependencies": {
    "cookie-parser": "~1.4.4",
    "debug": "~2.6.9",
    "express": "^4.21.2",
    "fs": "^0.0.1-security",
    "leaflet": "^1.9.4",
    "leaflet-gpx": "^2.2.0",
    "morgan": "~1.9.1"
    }

    But these dependencies can still be installed through the package
    manager, if shipped there, can't they?
    --
    Nuno Silva
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc,comp.lang.python on Mon Mar 9 13:59:42 2026
    From Newsgroup: comp.os.linux.misc

    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    If you actually want someone to help solve your problem then you’ll have
    to accurately describe what you are trying and what is happening. In
    practice this means posting the exact code you are using and exactly
    what happens when you run it.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Charlie Gibbs@[email protected] to comp.os.linux.misc on Mon Mar 9 16:34:52 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-09, rbowman <[email protected]> wrote:

    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it in
    C). It made me think that I could write something that my wife might
    actually be able to use. (I was careful to use Python 3, not 2.)
    It would be nice if this remained a viable option; otherwise it's
    back to writing Yet Another C Program.

    Stick to tkinter and you *should* be good. I say should because 'import tkinter' works on Mint, Ubuntu, and Fedora. On Leap you'd have to do
    'zypper install python-tk' but it would then be global. Arch says it's python isn't configured for Tk.

    My little test program uses tkinter, and runs fine on Debian Bookworm.
    Looks like a good start...
    --
    /~\ Charlie Gibbs | Growth for the sake of
    \ / <[email protected]d> | growth is the ideology
    X I'm really at ac.dekanfrus | of the cancer cell.
    / \ if you read it the right way. | -- Edward Abbey
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Mon Mar 9 22:17:07 2026
    From Newsgroup: comp.os.linux.misc

    On Mon, 09 Mar 2026 16:34:52 GMT, Charlie Gibbs wrote:

    On 2026-03-09, rbowman <[email protected]> wrote:

    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it in
    C). It made me think that I could write something that my wife might
    actually be able to use. (I was careful to use Python 3, not 2.)
    It would be nice if this remained a viable option; otherwise it's back
    to writing Yet Another C Program.

    Stick to tkinter and you *should* be good. I say should because 'import
    tkinter' works on Mint, Ubuntu, and Fedora. On Leap you'd have to do
    'zypper install python-tk' but it would then be global. Arch says it's
    python isn't configured for Tk.

    My little test program uses tkinter, and runs fine on Debian Bookworm.
    Looks like a good start...

    https://www.i-programmer.info/news/216-python/18680-microsoft-unifies- python-environments.html

    I haven't tried the Python Environments extension in VS Code yet and post
    this mainly for the xkcd cartoon that I didn't find a direct link for.

    The extension arrives a little late for me, having been through the
    degraded Python phase several years ago. Lessons learned:

    Stick with the Python installed by the distro. It's what the distro uses behind the scenes.

    Many Python projects are well within the capabilities of the Python itself
    and the system site-packages. No problem, life goes on as before.

    Some additional packages can be installed from the distro repositories
    with something like 'sudo apt install python3-foo'. In that case, no
    problem, and life goes on...

    If you want to use PyQt, PySide6, PyGtk, PyTorch, Numpy, and so forth they probably will not be in the distros repositories. Create a venv, activate
    it, and then use pip to install them. Passing --system-site-packages to
    venv may or may not been needed. Use 'pip freeze -l' to save a list of
    what you have installed so you can recreate the venv.

    Use venv, not virtualenv, conda, or some other scheme. Do not try to game
    the setup by tweaking environment variables.

    Probably not necessary if you play by the rules but i use 'python3 -m
    venv' and 'python3 -m pip' If you've created the xkcd cartoon with
    'another pip???' that makes sure that python and that pip are on the same page.

    Really, it isn't hard.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Tue Mar 10 02:38:48 2026
    From Newsgroup: comp.os.linux.misc

    On 3/9/26 12:34, Charlie Gibbs wrote:
    On 2026-03-09, rbowman <[email protected]> wrote:

    On Mon, 09 Mar 2026 02:47:51 GMT, Charlie Gibbs wrote:

    I did write a little Python program that played with the various GUI
    gadgets that it made available (with much less effort than doing it in
    C). It made me think that I could write something that my wife might
    actually be able to use. (I was careful to use Python 3, not 2.)
    It would be nice if this remained a viable option; otherwise it's
    back to writing Yet Another C Program.

    Stick to tkinter and you *should* be good. I say should because 'import
    tkinter' works on Mint, Ubuntu, and Fedora. On Leap you'd have to do
    'zypper install python-tk' but it would then be global. Arch says it's
    python isn't configured for Tk.

    My little test program uses tkinter, and runs fine on Debian Bookworm.
    Looks like a good start...

    TK can do a LOT of stuff ... more than good enough
    for complex touchscreen apps and such. User-driven
    and timer-driven ops are possible.

    TK is a little more plain and simple than GTK/QT
    "solutions" and much better documented. If I need
    a Python GUI I'll still use TK.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 02:49:19 2026
    From Newsgroup: comp.os.linux.misc

    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean
    and easy pre-VENV now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine'
    unless you PLAN to isolate the base machine from
    the application to various degrees.

    If you actually want someone to help solve your problem then you’ll have
    to accurately describe what you are trying and what is happening. In
    practice this means posting the exact code you are using and exactly
    what happens when you run it.

    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    Or do I have to go back to 2020 versions of Linux ?

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 08:47:25 2026
    From Newsgroup: comp.os.linux.misc

    c186282 <[email protected]> writes:
    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    You’ve not reported anything actionable. Some stuff doesn’t work from
    your script, well so what? Without seeing concrete details, nobody can
    tell whether there’s a real behavior change there or just a bug in your script. Since it all works fine for everyone else, the working
    assumption is the latter.

    I want the VENV *dead* ... CAN that be done ???

    The answer to that has already been posted: if you are in a venv and
    don’t want to be, just type ‘deactivate’.

    Or do I have to go back to 2020 versions of Linux ?

    venvs have been around for something like 15 years, so if the problem
    was really venvs then dialing back to 2020 is unlikely to make a
    difference.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 06:14:25 2026
    From Newsgroup: comp.os.linux.misc

    On 3/10/26 04:47, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    You’ve not reported anything actionable.
    Then don't waste your time.

    A report is a report.

    If others experience the same issues maybe
    they will go deeper into it, or not.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Tue Mar 10 06:21:17 2026
    From Newsgroup: comp.os.linux.misc

    On 3/9/26 04:41, rbowman wrote:
    On Sun, 8 Mar 2026 21:14:00 -0400, c186282 wrote:

    At some point the 'idealists' can degrade and ruin a perfectly good
    thing. The idea of a Python VE is a stone on this road to ruin.

    I disagree. VEs save you from the equivalent of dll hell someplace down
    the road. Dotnet, Angular, Node, and others are similar now. Your project encapsulates the dependencies it needs. Another project may require
    different versions. For example with node anything you install for the project goes into node_modules and package.json has the dependencies for
    that project:

    "dependencies": {
    "cookie-parser": "~1.4.4",
    "debug": "~2.6.9",
    "express": "^4.21.2",
    "fs": "^0.0.1-security",
    "leaflet": "^1.9.4",
    "leaflet-gpx": "^2.2.0",
    "morgan": "~1.9.1"
    }

    Hmmm ... Python worked GREAT *before* the VENV ...
    no 'hell'.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Rich@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 14:55:34 2026
    From Newsgroup: comp.os.linux.misc

    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean
    and easy pre-VENV now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine'
    unless you PLAN to isolate the base machine from
    the application to various degrees.

    If you actually want someone to help solve your problem then you’ll have >> to accurately describe what you are trying and what is happening. In
    practice this means posting the exact code you are using and exactly
    what happens when you run it.

    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    And, again, much as Jon said, you are blaming "venv's" but given us no examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python than
    you do, something to analyze and maybe show you where the problem
    really exists.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Pancho@[email protected] to comp.os.linux.misc on Tue Mar 10 15:02:03 2026
    From Newsgroup: comp.os.linux.misc

    On 3/9/26 08:41, rbowman wrote:
    On Sun, 8 Mar 2026 21:14:00 -0400, c186282 wrote:

    At some point the 'idealists' can degrade and ruin a perfectly good
    thing. The idea of a Python VE is a stone on this road to ruin.

    I disagree. VEs save you from the equivalent of dll hell someplace down
    the road.
    +1.

    I hate to think how much time I have wasted on environment. Many
    projects I worked on took days fix configuration when onboarding a new programmer.

    Even a cursory read of this thread suggests the reason the venv doesn't
    work is because of a poorly understood environment.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 13:03:49 2026
    From Newsgroup: comp.os.linux.misc

    On 3/10/26 10:55, Rich wrote:
    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean
    and easy pre-VENV now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine'
    unless you PLAN to isolate the base machine from
    the application to various degrees.

    If you actually want someone to help solve your problem then you’ll have >>> to accurately describe what you are trying and what is happening. In
    practice this means posting the exact code you are using and exactly
    what happens when you run it.

    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    And, again, much as Jon said, you are blaming "venv's" but given us no examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python than
    you do, something to analyze and maybe show you where the problem
    really exists.

    Can't figure out how to zap the VENV in order
    to provide the examples you'd like :-)

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Jon Ribbens@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 17:09:21 2026
    From Newsgroup: comp.os.linux.misc

    On 2026-03-10, c186282 <[email protected]> wrote:
    On 3/10/26 10:55, Rich wrote:
    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean
    and easy pre-VENV now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine'
    unless you PLAN to isolate the base machine from
    the application to various degrees.

    If you actually want someone to help solve your problem then you’ll have >>>> to accurately describe what you are trying and what is happening. In
    practice this means posting the exact code you are using and exactly
    what happens when you run it.

    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    And, again, much as Jon said, you are blaming "venv's" but given us no
    examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python than
    you do, something to analyze and maybe show you where the problem
    really exists.

    Can't figure out how to zap the VENV in order
    to provide the examples you'd like :-)

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.

    This not only isn't true, it cannot be true, because *by definition*
    if the environment you're seeing is the same one Python always sees
    then it isn't a "virtual" environment, it's the system environment.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 17:56:03 2026
    From Newsgroup: comp.os.linux.misc

    c186282 <[email protected]> writes:
    On 3/10/26 10:55, Rich wrote:
    And, again, much as Jon said, you are blaming "venv's" but given us
    no examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python
    than you do, something to analyze and maybe show you where the
    problem really exists.

    To do that they would have to want their problem to be solved, rather
    than to have something external to blame for it.

    Can't figure out how to zap the VENV in order
    to provide the examples you'd like :-)

    You’ve already been told how to exit a venv twice.

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.

    It is not.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Rich@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 18:13:38 2026
    From Newsgroup: comp.os.linux.misc

    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/10/26 10:55, Rich wrote:
    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean and easy pre-VENV
    now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine' unless you
    PLAN to isolate the base machine from the application to various
    degrees.

    If you actually want someone to help solve your problem then
    you’ll have to accurately describe what you are trying and what is
    happening. In practice this means posting the exact code you are
    using and exactly what happens when you run it.

    Too fuckin' old now, not gonna get into a million details and
    sub-sub-sub-config-files. I just report that some lower-level
    OS commands no longer work as expected, or at all, using the
    usual Python utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    And, again, much as Jon said, you are blaming "venv's" but given us
    no examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python
    than you do, something to analyze and maybe show you where the
    problem really exists.

    Can't figure out how to zap the VENV in order to provide the
    examples you'd like :-)

    In that case, how about a very simple example that fails, for you, so
    the others here (Jon, etc.) can show you what you screwed up?

    I predict no such example will be posted, since you seem to want to
    bitch and complain about things you don't want to understand more than
    you want to know why your system call isn't working.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From piergiorgio.sartor.this.should.not.be.used@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 19:14:26 2026
    From Newsgroup: comp.os.linux.misc

    On 10/03/2026 18.03, c186282 wrote:
    [...]
      Can't figure out how to zap the VENV in order
      to provide the examples you'd like  :-)

    There are the following possibilities here:
    1) You didn't properly get what a venv is in Python
    2) You're trolling.

    Assuming 1), the venv is nothing more that a
    search path change, for Python programs and
    modules. Where, instead of looking first into
    the system folders, first is searched in the
    venv "repository" area.
    Nothing more, nothing less.

    So, what do you want to "zap"?
    There is nothing to "zap"...
    Unless for "zapping" you meaning something
    different. In case, please specify.

      Seems like, of late, it's entirely baked-in -
      Python is the VENV and the VENV is Python.

    Non clear what this means, but maybe is the
    same written above.

    bye,
    --

    piergiorgio
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc on Tue Mar 10 18:24:40 2026
    From Newsgroup: comp.os.linux.misc

    On Tue, 10 Mar 2026 06:21:17 -0400, c186282 wrote:

    Hmmm ... Python worked GREAT *before* the VENV ...
    no 'hell'.

    Until it didn't. The install of the Esri desktop included a Python2.7 directory that had Python with their secret sauce added. It did not play
    well if you had Python from python.org installed. When they finally
    migrated to Python 3 they used conda for the installation. That REALLY
    didn't play well.

    Then there was ActiveState Python, originally designed for Windows with packages to leverage the Windows API. They also had a Perl installation.

    Sure, you could use Python casually and not run into any problems. You
    still can if you don't want packages that aren't included in the distro's site-packages.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 18:26:58 2026
    From Newsgroup: comp.os.linux.misc

    On Tue, 10 Mar 2026 02:49:19 -0400, c186282 wrote:

    Or do I have to go back to 2020 versions of Linux ?

    You can, but you will have to learn how to use uv and you'll bitch about
    that too.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 19:36:23 2026
    From Newsgroup: comp.os.linux.misc

    On Tue, 10 Mar 2026 13:03:49 -0400, c186282 wrote:


    Can't figure out how to zap the VENV in order to provide the examples
    you'd like

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.

    Unless you've specifically created a venv, you're not in one. I have a
    script that scrapes the current conditions and forecasts from the NOAA
    site. It uses two imports.

    import json
    import requests

    json is built in, requests is in the dist-packages.

    $ python3 -m site
    sys.path = [
    '/',
    '/usr/lib/python313.zip',
    '/usr/lib/python3.13',
    '/usr/lib/python3.13/lib-dynload',
    '/usr/local/lib/python3.13/dist-packages',
    '/usr/local/lib/python3.13/dist-packages/opensky_api-1.3.0-
    py3.13.egg',
    '/usr/lib/python3/dist-packages',
    ]

    in this case 'ls /usr/lib/python3/dist-packages' will show you what is available. Distros install many as a default. Others can be added like

    'sudo apt install python3-numpy'

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc,comp.lang.python on Tue Mar 10 23:51:59 2026
    From Newsgroup: comp.os.linux.misc

    On Tue, 10 Mar 2026 18:13:38 -0000 (UTC), Rich wrote:

    I predict no such example will be posted, since you seem to want to
    bitch and complain about things you don't want to understand more than
    you want to know why your system call isn't working.

    FFmpeg is all screwed up in Debian Trixie
    I can't install VirtualBox in Fedora.
    Adding kvm to the kernel by default screwed everything up.
    Linux Mint is hosed, too.
    Got a VM up but ffmpeg still doesn't work.
    Trying something with Python but that doesn't work.
    The venv I'm probably not in screwed everything up.
    Nah, I don't want to do that; it might work.

    I think I'm seeing a pattern here. I'll ask Claude.
    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc,comp.lang.python on Wed Mar 11 02:48:33 2026
    From Newsgroup: comp.os.linux.misc

    On 3/10/26 13:09, Jon Ribbens wrote:
    On 2026-03-10, c186282 <[email protected]> wrote:
    On 3/10/26 10:55, Rich wrote:
    In comp.os.linux.misc c186282 <[email protected]> wrote:
    On 3/9/26 09:59, Richard Kettlewell wrote:
    c186282 <[email protected]> writes:
    The Python VENV *seems* to be thwarting my
    ability to run certain hard-core system
    routines via "os.system()" or related. The
    'virtual' seems to be for idiot 7-year-olds
    you don't want to crash the box.

    Well, sometimes, I *want* to crash the box.

    As Jon says, you’ve misdiagnosed your problem.

    COULD be ...

    But I still note that things that worked clean
    and easy pre-VENV now DON'T.

    Who/what should I blame ?

    There's no POINT in a 'virtual environment/machine'
    unless you PLAN to isolate the base machine from
    the application to various degrees.

    If you actually want someone to help solve your problem then you’ll have
    to accurately describe what you are trying and what is happening. In >>>>> practice this means posting the exact code you are using and exactly >>>>> what happens when you run it.

    Too fuckin' old now, not gonna get into a million
    details and sub-sub-sub-config-files. I just report
    that some lower-level OS commands no longer work
    as expected, or at all, using the usual Python
    utilities for doing so.

    I want the VENV *dead* ... CAN that be done ???

    And, again, much as Jon said, you are blaming "venv's" but given us no
    examples to support your assertion. It is *very likely* that you
    indeed are blaming the wrong thing. How about creating a very short
    example that works without a venv but fails within one, and posting
    that example. That will give others, who may know far more python than
    you do, something to analyze and maybe show you where the problem
    really exists.

    Can't figure out how to zap the VENV in order
    to provide the examples you'd like :-)

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.

    This not only isn't true, it cannot be true, because *by definition*
    if the environment you're seeing is the same one Python always sees
    then it isn't a "virtual" environment, it's the system environment.


    OK, tomorrow, HOW do I start a Py3 that's NOT in
    the VENV ???

    Go ahead, search - I have. All the 'fixes' don't
    work anymore.

    Many say "no problem" - yet NO current workarounds.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From rbowman@[email protected] to comp.os.linux.misc,comp.lang.python on Wed Mar 11 07:32:17 2026
    From Newsgroup: comp.os.linux.misc

    On Wed, 11 Mar 2026 02:48:33 -0400, c186282 wrote:


    OK, tomorrow, HOW do I start a Py3 that's NOT in the VENV ???

    Go ahead, search - I have. All the 'fixes' don't work anymore.

    Many say "no problem" - yet NO current workarounds.

    What the fuck are you trying to do?

    $ python3
    Python 3.13.7 (main, Jan 22 2026, 20:15:57) [GCC 15.2.0] on linux
    Type "help", "copyright", "credits" or "license" for more information.


    At the command prompt you can do interactive.

    python3 foo,py

    runs foo.py.

    I finally found a use for the caps lock.

    YOU'RE NOT IN A FUCKING VENV UNLESS YOU CREATE IT, GO INTO THE DIRECTORY
    AND RUN '. bin/activate

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From c186282@[email protected] to comp.os.linux.misc on Wed Mar 11 04:20:07 2026
    From Newsgroup: comp.os.linux.misc

    On 3/10/26 14:24, rbowman wrote:
    On Tue, 10 Mar 2026 06:21:17 -0400, c186282 wrote:

    Hmmm ... Python worked GREAT *before* the VENV ...
    no 'hell'.

    Until it didn't. The install of the Esri desktop included a Python2.7 directory that had Python with their secret sauce added. It did not play
    well if you had Python from python.org installed. When they finally
    migrated to Python 3 they used conda for the installation. That REALLY
    didn't play well.

    Then there was ActiveState Python, originally designed for Windows with packages to leverage the Windows API. They also had a Perl installation.

    Sure, you could use Python casually and not run into any problems. You
    still can if you don't want packages that aren't included in the distro's site-packages.

    Well, 2.7 was a LONG time ago. I'm talking
    the latest 3.x

    Hmmm ... maybe I need to install Py2 ? :-)

    Keep hearing "no prob" - yet NO handy solutions
    for starting/running Py3 outside the VENV.

    Oh well, starting on some Free Pascal alts
    for some of my Py apps. Has to be done - and
    I don't mind Pascal one bit though Python IS
    more friendly with slicing strings and such.

    Oh, weird, that ffmpeg prob that started all
    these threads ... still exists in F43. You
    can only start it SO many times, then it
    hangs or won't start right. SOME error msgs
    about a 'transport' error - maybe ffmpeg
    loses track of the ip cam for a few secs
    every so often. Tried a few suggested buffer
    fixes ... don't work. And, oddly, the prob
    remains most prominent during certain hours
    of the day, like someone screwed up some
    time/date code within ffmpeg.

    Best current 'fix' - start ffmpeg as an
    independent process, just add a "&" to the
    command line, and a sleep statement for a
    little longer than it's supposed to run.
    Finally - os.execv() to restart the pgm
    in-place beats internal looping or letting
    it drop with systemd restarting.

    Oh, 'os.execv()' ... the docs do NOT really
    describe how to evoke it ... the syntax is
    very weird. Seek working examples online.

    Attempting 'os.exec("sudo killall -9 ffmpeg")'
    after each cycle, but that seems to be one of
    the system things the VENV won't pass properly.
    Works perfectly from a real CL, but not from
    within the Py app.

    --- Synchronet 3.21d-Linux NewsLink 1.2
  • From Richard Kettlewell@[email protected] to comp.os.linux.misc,comp.lang.python on Wed Mar 11 08:40:36 2026
    From Newsgroup: comp.os.linux.misc

    c186282 <[email protected]> writes:
    On 3/10/26 13:09, Jon Ribbens wrote:
    On 2026-03-10, c186282 <[email protected]> wrote:
    Can't figure out how to zap the VENV in order
    to provide the examples you'd like :-)

    Seems like, of late, it's entirely baked-in -
    Python is the VENV and the VENV is Python.
    This not only isn't true, it cannot be true, because *by definition*
    if the environment you're seeing is the same one Python always sees
    then it isn't a "virtual" environment, it's the system environment.

    OK, tomorrow, HOW do I start a Py3 that's NOT in
    the VENV ???

    You’ve been told how to exit venvs more than once. If you can’t follow simple instructions, that’s a problem with you, not Python.

    Go ahead, search - I have. All the 'fixes' don't
    work anymore.

    Many say "no problem" - yet NO current workarounds.

    Multiple people are trying to help you, for increasingly opaque
    reasons. You’ll get nowhere until you actually start following their
    advice.
    --
    https://www.greenend.org.uk/rjk/
    --- Synchronet 3.21d-Linux NewsLink 1.2