cfbolz changed the topic of #pypy to: PyPy, the flexible snake (IRC logs: https://quodlibet.duckdns.org/irc/pypy/latest.log.html#irc-end ) | use cffi for calling C | if a pep adds a mere 25-30 [C-API] functions or so, it's a drop in the ocean (cough) - Armin
<StyXman>
arigato: I don't see the «git repo with three commits» image
<StyXman>
is tehre one?
<arigato>
if you want me to actually make an *image* out of the three-commits explanation of why we don't want Git, because you think people won't read that, then I honestly don't care
<mattip>
arigato: the wording is a bit misleading "Look at this git repo" seems to imply there is a git repo to look at: either an image or a set of commands to recreate it
<arigato>
ah OK
<mattip>
perhaps "consider a git repo"
<arigato>
thanks
<arigato>
StyXman: thanks! sorry if I feel a little bit aggressive here. It's really trying to answer an "Extremely Frequently Asked Question", so it makes sense to try to do it right
<StyXman>
hmm, I'm a visual guy, lemme try again
<StyXman>
ok, I got it. I would also add the text «hint: it could be neither»
<StyXman>
(the commit could have been done on a branch whose head was #1 but has since been deleted)
<arigato>
ah, you can even delete git branches so that they are completely gone? I didn't know
<cfbolz>
StyXman: thanks for helping make this clearer. Sorry that we're a bit tetchy, discussing version control systems is always a bit tiring
<StyXman>
hehehe, I can understand
<arigato>
yes
<arigato>
but thanks! Improved further now. Will show up in a few minutes on doc.pypy.org
<mattip>
it does pretty well with the content, but ignores the comments
astronavt has quit [Quit: ...]
astronavt has joined #pypy
xcm is now known as Guest59374
Guest59374 has quit [Killed (weber.freenode.net (Nickname regained by services))]
xcm has joined #pypy
<antocuni>
mattip: I don't think we have many interesting comments
<mattip>
heh
<LarstiQ>
arigato: in fact, the git style is to aggressively delete branches after they are merged, all the relevant information is considered to be in the revision dag itself
<Dejan>
122k people are watching this rocket launch
<LarstiQ>
(and from that point of view the "Why doesn’t PyPy use Git and move to GitHub?" FAQ does not explain why being able to answer the question ` When commit #1 was made, was it in the branch “A” or “B”?` is of value)
<LarstiQ>
perhaps leaving some people puzzled
<Dejan>
I understand that PyPy uses mercurial's named branches, and there is no such thing in Git
<mattip>
LarstiQ: but it is really a different way of looking at a chain of changesets
<LarstiQ>
mattip: I'm afraid I don't understand myself what the value is, and I don't want to aggravate you all by trying to interview it out of you
<LarstiQ>
so here was just trying to explain why people might not be convinced by the FAQ answer
<Dejan>
I just wish there was git-hg just like there is git-svn :)
<mattip>
the idea is that a single changeset from "git blame" does not give enough perspective on what was done and why,
<mattip>
you need to look at the entire branch, where it was in relation to other branches
<mattip>
and what it built on
<Dejan>
mattip, i wonder how often you reall need that info
<Dejan>
s/reall/really/
<mattip>
well, that is the difference. Git people never had the info, so they don't miss it.
<LarstiQ>
learning how this information is used to answer a question/solve a problem would maybe be instructive, atm I'm still wondering how that's different from looking at `git log --graph` (maybe with `--all`)
<mattip>
after a few commits, the graph gives up and draws a disconnected line with an arrow
<mattip>
now if the branch has dozens of commits, and you move around to figure out what happenned in relation to other branches, how do you know which branch is which?
<mattip>
in hg, it is part of the first-order information
<mattip>
given a random commit in git, what branch is it on?
<antocuni>
mattip: one option is to import the old blog posts AND to provide a link to the original one
<nedbat>
mattip: you can ask which existing branches have the commit in their history, but that isn't the same as having the branch name baked into the commit as hg had it.
<mattip>
example: someone successfully imported pypy into github. What branch was this commit on from 2011?
<Dejan>
that is all nice, but how often do you really use it?
<Dejan>
I have ~40 devs here and we needed something like what you are describing once a year
<Dejan>
in fact i can't remember when was the last time we were actually looking after the branch in which particular commit was made
<Dejan>
even git blame usage is rare
<antocuni>
Dejan: it is essential to know the context in which an obscure line was added. The once a year time you need it, it saves your day
<nedbat>
mattip: i can't tell why you are asking: 1) i want to learn how to get whatever info i can from git, 2) i want to stay on hg, 3) I'm annoyed. No judgement. :)
<antocuni>
and hg blame, I use it every time. So this probably means that pypy has different characteristics than the projects you work on :)
<mattip>
nedbat: I want to stay on hg, and it makes me sad other people reject it out of hand
<nedbat>
mattip: is staying on hg an option?
<Dejan>
nedbat, i thought that is settled
<mattip>
yup, that's why we moved from bitbucket to foss.heptapod.net
<nedbat>
Dejan: idk, i'm asking
<mattip>
but then people started asking why we like mercurial, so I am trying to explain
<Dejan>
As old CSV/SVN user I like Mercurials commit numbers :)
<gracinet>
I wouldn't mind them handing us a few tens of k$ :-)
<gracinet>
mattip: yes, there are thousands of places like this
<mattip>
gracinet: "gitlab + mercurial" or something snappy like that
<mattip>
it could be in the header
<antocuni>
hgitlab :)
<mattip>
or on the icon hover
<gracinet>
I'd prefer to tackle them once we are on a more current version of GitLab (what I'm doing right now) - up to now, it's been "those that are the most disturbing only"
<mattip>
cool
<tos9>
antocuni: careful, along with hpy, adding "H" in front of things is going to become your signature move
antocuni is now known as hantocuni
<mattip>
heh
<tos9>
:)
hantocuni is now known as antocuni
<gracinet>
Actually, these cosmetic things are also some I keep to encourage contributors
<gracinet>
mattip: ignore request access done as bjones14 (that was me testing with first user I could find)
<gracinet>
I'm going to make a permanent testgr user, it'll be better
<LarstiQ>
mattip: "now if the branch has dozens of commits, and you move around to figure out what happenned in relation to other branches, how do you know which branch is which?" -- I look at the revision graph, the name of the branch doesn't matter? It seems clear to me you have a very particular usage in mind that I'm not aware of, so it's very hard to communicate. I absolutely believe you have a use, I'm just
<LarstiQ>
missing the context.
bitbit has joined #pypy
<LarstiQ>
Perhaps the answer lies along: long-lived branches that criss-cross?
<mattip>
yes. If you have never driven a truck it will be hard to explain exactly how important those little posts that stick up over the hood are.
<mattip>
but once you try to sit in the driver seat, you realize you cannot live without them.
* gracinet
does not have a driver's license
<mattip>
this is the same kind of thing. We have a workflow that works for us
<mattip>
others don't see the need and some tell us "get over it, everyone works with git". We don't want to.
<mattip>
there is another tool, it is very capable, why don't you just use it instead?
<mattip>
it kind of reflects the "CPython + cython/numba" vs. pypy argument. People are happy with CPython, it works for them
<mattip>
that's great
<LarstiQ>
mattip: No argument from me to switch to git, just trying to understand as long as you're willing to indulge me, and trying to help by getting an explanation that will forver shut down these queries
CrazyPython has joined #pypy
bitbit has quit [Ping timeout: 268 seconds]
bitbit has joined #pypy
<bitbit>
Hi. I did "pypy3 -m ensurepip --user" so that I can use pip3 with pypy3, but now when I use python3 it doesn't have the module I installed with pip3. is there a way to use pip with pypy and python in parallel?
<mattip>
no. while pip may appear to be something you run from the command line, it actually is not
CrazyPython has quit [Read error: Connection reset by peer]
plan_rich has joined #pypy
<plan_rich>
mattip: hi, well that is awkward. The creation of the virtualenv fails because using the pypy executable it cannot determine the python version. Looking at the code of virtualenv right now...
slavfox has joined #pypy
bitbit has quit [Remote host closed the connection]