On Mon, Nov 07, 2022 at 02:53:55PM -0500, Stewart Russell via talk wrote:
You bet! I think it will continue for some time.
I *think* that systems are only supposed to provide an executable called "python3" now, but de facto, the executable "python" seems to be a free-for-all. I only found one current system of mine where /usr/bin/python was Python 2.7: somehow I'd got python-is-python2 installed there.
If python 2.7 is installed, it is correct for it to be named /usr/bin/python. If python 2 is not installed, /usr/bin/python should not exist at all. Python 3 should always be /usr/bin/python3 and hence no conflict can ever exist between them.
C++ never said it was going to replace C entirely, though. And the commands to invoke the compilers have always been different.
Yes C++ never claimed to be C.
Yes: you can leave things in a mess doing that. So far I've never broken things entirely. So far.
I occasionally have to check that my system Python hasn't been hijacked by a large package install. Anaconda did that to me once, putting its own path ahead of my one without telling me. Conversely, many of the embedded build systems I use hide their own Python3 distributions deep in their own path so they can guarantee that it'll work. Some of these are so huge (Espressif ESP IDF 4.4 comes in at around 20 GB with all dependencies to build on Tensilica and RISC-V cores) that an entire Python distro is a tiny part of the whole.
The size of TeX Live makes me sad, because I remember it being "unimaginably complete" when my late and vastly effusive friend Sebastian Rahtz announced the project as filling a whole CD-ROM in the late 1990s. It's now many times that size, and will automatically be installed on your system if you go near any Gnu documentation.
I only use one packaging system on my machines. pip and anaconda are evil abomonations and should never be used. :) I despise how many new languages think it's their job to invent a new distribution platform for modules for that language. -- Len Sorensen