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
<antocuni>
061e5c3b9d32, 2be8d722c87c, ed6c9b79a62c and 58ecc4889d11
<antocuni>
is it "intended" or it is an artifact of the heptapod importing?
<ronan>
antocuni: it's all intended, I guess. 58ecc4889d11 is the last Bitbucket commit, adding the move notice. 2be8d722c87c, ed6c9b79a62c have topics. 061e5c3b9d32 is the real head
<mattip>
I guess we should merge 58ecc4889d11 into 061e5c3b (but ignore the changes) to reduce chances of a problem
<mattip>
antocuni: I cleaned up the bitbucket head: closed and merged
<antocuni>
ronan, mattip: I suppose I should learn what are "topics" then :)
<antocuni>
are they basically braches, with their own head, but they are supposed to disappear eventually?
<antocuni>
mattip: thanks
<mattip>
yes, they are git branches: when merged they will loose their identity
<ronan>
antocuni: Yes, you need to install hg-evolve and add 'evolve =' and 'topic =' in your .hgrc to see them
<ronan>
we should probably close these MRs though, they're completely obsolete
<mattip>
the problem, as you discovered, is that mercurial-without-evolve sees them as heads on the branch they came from
<mattip>
ronan: +1
<marmoute>
mattip: byt the way. This will be fixed in the next version. We have working code for this that just need some polish.
<mattip>
marmoute: +1
marky1991 has quit [Remote host closed the connection]
marky1991 has joined #pypy
<antocuni>
speaking of installing hg-evolve: how do you install mercurial? Do you use the one shipped by ubuntu, use pip install --user, use a virtualenv or whatever?
<antocuni>
ok, now hg no longer complains about new heads, but instead it says "4 new obsolescence markers"
<antocuni>
what the hell is it? :)
<mattip>
marmoute: ^^^
<marmoute>
antocuni: a small bit of information that allow for safe and distributed history rewriting
<marmoute>
`hg help evolution`
<marmoute>
(could probably use a refresh)
<antocuni>
thanks
<marmoute>
(also see point above where we won't make it necessary to have all this)
<arigato>
marmoute: maybe someone should focus on fixing the warnings that hg emits in parenthesis, because at least for our working style they are most often completely off
<arigato>
should I report these when I see them?
<marmoute>
arigato: which one ?
<marmoute>
antocuni: I usually install mercurial from package (but sometime from pip) and evolve using pip --user
<antocuni>
marmoute: ok cool, I was worried of weird interaction between versions or so
<marmoute>
the extensions is compatible with a range of Mercurial revision
<marmoute>
And it use namespace package to make the import story "simple"
<Dejan>
isn't one of the mercurials selling points that you can't rewrite history?
oberstet has quit [Remote host closed the connection]
<marmoute>
Dejan: you cannot unsafely do so :-p
oberstet has joined #pypy
plan_rich has joined #pypy
<arigato>
marmoute: for example, hg pull => (run 'hg update' to get a working copy) even if there are no new changes on the currently checked out branch
<marmoute>
yeah, it is easy to argue that this one is buggy
<arigato>
there are also other cases I can't think of right now which leads to a hint that's really buggy instead of just unnecessary like the previous one
<arigato>
maybe if 'hg pull' gives you a 2nd head (on the same branch you're now), but you're not on the branch tip already, it tells you about running 'hg merge' without 'hg up' first
<arigato>
and well, there's (branches are permanent and global, did you want a bookmark?) which I'm pretty sure I don't want
<marmoute>
I think we stop issued this one if the repository has named branch already
<arigato>
ah ok
<ronan>
hrm, if you close a named branch before merging, it automatically closes the merge request
<ronan>
marmoute: what are we supposed to do when merging a named branch? The MR doesn't seem to handle closing the branch
<marmoute>
ronan: hum Gitlab have UI to deal with this kind of things. But we probably did not plug them to named branch yet.
<ronan>
marmoute: well, there's a disabled box labeled "Remove source branch", which sounds like a git concept
<marmoute>
We could make it "close source named branch"
<ronan>
marmoute: +1, that's what Bitbucket does
<marmoute>
ronan: can you file a ticket so that I can keep track of it ?
<arigato>
mattip: it's really a problem triggered by cffi
<arigato>
I have no clue why it fails here but it works on azure when I release cffi for cpython
<arigato>
maybe because there is a package missing on this builder
<arigato>
but who knows
marky1991 has joined #pypy
oberstet has quit [Remote host closed the connection]
<mattip>
arigato: it was fine until the buildslave updated to a the latest macOS with a later xcode
<mattip>
it didn't even include a libffi, they had to install from homebrew
<mjacob_>
arigato: how's the weather in leysin?
<arigato>
mattip: OK, I'll stand by my "I have no clue"
<arigato>
you should see the exact command lines run in the logs, and these sound reasonable to me, so my guess is that clang gives this output because of a misconfiguration
<arigato>
I doubt that adding "-stdlib=libc++" would magically solve these issues
<arigato>
mjacob_: from tomorrow on, nice weather. today and tonight, bad
CrazyPython has joined #pypy
<arigato>
well, more precisely, they announce one nice day and one bad day alternating a few times
<mattip>
the log seems to indicate it is compiling with cc and linking with c++
<mattip>
after generating _CFFI_test_math_sin.cpp
xcm has quit [Remote host closed the connection]
<arigato>
it's the same on Linux, I think gcc is able to detect that the extension is .cpp and call g++ itself
<mattip>
maybe clang cannot do that
<arigato>
yes, gcc definitely do that
xcm has joined #pypy
<arigato>
no, actually clang too, because the file .cpp contains ``extern "C" { .. }``
<arigato>
which doesn't make sense as C
<mattip>
maybe clang only goes half way and still needs the flag, like it says in the warning
<mattip>
google is not being helpful
<arigato>
I still think it's a misconfiguration
<mattip>
I found a comment somewhere about " libcxx and clang are separate packages, so even if clang has the same fallback it won't find anything on it's own"
<mattip>
so maybe they needs to install libcxx?
<mattip>
tos9 ( or any other macOS user ): ping
CrazyPython has quit [Read error: Connection reset by peer]
CrazyPython has joined #pypy
jacob22 has quit [Read error: Connection reset by peer]
<tos9>
mattip: semipong
CrazyPython has quit [Ping timeout: 258 seconds]
Smigwell has left #pypy [#pypy]
jacob22 has joined #pypy
<mattip>
tos9: if you could take a look at the discussion from a few hours ago and look at the build log linked there,
<mattip>
I don't understand the clang error message, and am not having alot of luck piecing together the puzzle by looking at internet search results
xcm has quit [Read error: Connection reset by peer]