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
shunning has quit [Remote host closed the connection]
<kenaan>
Yannick_Jadoul py3.7-bpo-31333 4ad607fdca5d /pypy/: Not importing _weakrefset anymore to implement _abc module, and added '_abc' to essential_modules ...
<kenaan>
Yannick_Jadoul py3.7-bpo-31333 8c14c8098e5a /pypy/module/_abc/app_abc.py: Fixing _abc._get_dump to return sets, rather than the new SimpleWeakSet wrapper
<kenaan>
Yannick_Jadoul py3.7-bpo-31333 1e2d3e37d725 /pypy/module/_abc/: Added tests and fixed _reset_registry and _reset_caches
<kenaan>
Yannick_Jadoul py3.7-bpo-31333 cf9fa90f1848 /pypy/config/pypyoption.py: Moving _abc from essential_modules to default_modules
<cfbolz>
YannickJadoul: if you want, you could have commit rights now. Then you could work on a branch in the main repo and still ask us for reviews
<arigato>
cfbolz: would you agree with this new error message?:
<arigato>
>>>> class A([]): pass
<arigato>
File "<inline>", line 1 in <module>
<arigato>
class A([]): pass
<arigato>
Traceback (application-level):
<arigato>
TypeError: metaclass found to be 'list', but calling <type 'list'> with args (name, bases, dict) raised 'list() takes at most 1 argument (3 given)'
<cfbolz>
arigato: nice!
<YannickJadoul>
cfbolz: Oh, nice :)
<YannickJadoul>
I don't really mind either way, but you all prefer that workflow?
<cfbolz>
arigato: maybe we can even use the real name and the real bases? The dict is impractical of course
<arigato>
ah yes
<cfbolz>
YannickJadoul: the PyPy approach is indeed to all work in the main repo
<arigato>
cfbolz: I tried to use the hack of only changing the error message if the current exception is, of course, a TypeError, but also if it doesn't have any traceback attached so far
<cfbolz>
Right
<arigato>
does it make any sense? and would it be useful in more cases?
<YannickJadoul>
cfbolz: OK, will do :) And then just create PRs as usual, or ask in IRC/somewhere else?
<cfbolz>
arigato: ah, no traceback is a nice extra idea. Yes, sounds sensible
<cfbolz>
YannickJadoul: yes, just ask here, or open a pull request if you want
<ronan>
YannickJadoul: also, that way you can kick buildbots on your branches
<cfbolz>
Right
<arigato>
OK cool
<cfbolz>
YannickJadoul: what's your bit bucket user name?
<YannickJadoul>
cfbolz: As you prefer; I don't mind. I tend to like writing up decisions I made and linking to the relevant CPython, but if that's just extra spam for you...
<YannickJadoul>
It's Yannick_Jadoul
<YannickJadoul>
(Very boringly :) )
<cfbolz>
YannickJadoul: that can still go in the merge commits (where it will remain visible even after we have to leave bit bucket)
<YannickJadoul>
cfbolz: Ah, yes, true. Though I don't make those if any of you would merge. But I'll remember to add them to the commit message myself, then
<arigato>
YannickJadoul: fwiw, our style is to prefer slightly messy commit histories and the more comments in the messages the better, rather than cleaned-up histories where the original trials of the person doing it might be lost
<YannickJadoul>
arigato: Ha, opposite to my habit of force-pushing the minor details that failed the CI on GitHub
<ronan>
YannickJadoul: I like opening PRs for branches, that saves me having to remember the correct procedure for merging, and it's nice to have a GUI for the review
<YannickJadoul>
But anyway, that's not possible with mercurial anyway
<cfbolz>
YannickJadoul: Yay, welcome to being a PyPy committer!
<cfbolz>
Keep up the good work!
<YannickJadoul>
Hehe, thanks! :) I'll wear this badge with pride ;)
<mattip>
antoniocuni added pypy/tool/release/smartstrip.py about two years ago. It uses objcopy. I am using it on libpypy.so produced by gcc8 on centos6
<mattip>
it errors out, "not enough room for program headers, try linking with -N"
<mattip>
anyone know what '-N' does?
<mattip>
this happened on the keep-debug-symbols branch 005844269621
<arigato>
'-N' is documented '...disable linking against shared libraries.'
<mattip>
--no-keep-debug used strip instead which failed to strip but succeeded to build the package
<mattip>
arigato: so it will link with the *.a libraries instead?
<arigato>
no, it will probably explode completely
<mattip>
I wonder why centos/gcc8 is linking differently
<arigato>
the error message sounds a lot like "the author of objcopy didn't put efforts in covering all cases, and in your case it doesn't work, sorry"
<mattip>
it happens when using strip as well, just the failure does not crash the packaging script
<kenaan>
mattip buildbot c40bc42d1f79 /docker/Dockerfile: add more documentation around using docker