These are chat archives for dropbox/pyston

21st
May 2016
lesshaste
@lesshaste
May 21 2016 09:28
Hi.. just a very very gentle reminder that a blog post update would be very gratefully received by your adoring fans :)
There is a risk of people seeing the "unladen swallow" otherwise :)
Sun
@Daetalus
May 21 2016 11:33
Agree
Marius Wachtler
@undingen
May 21 2016 15:41
we plan today a release next week so there should be a blog post soon :-)
Sun
@Daetalus
May 21 2016 15:42
And the resize function won't get fix?
Marius Wachtler
@undingen
May 21 2016 15:44
do you mean that we don't have the exact same refcounts, and therefore numpys resize errors sometimes? If it's that I don't think this is a release blocker.
Sun
@Daetalus
May 21 2016 15:45
Yeah, just the problem could not get fixed?
Marius Wachtler
@undingen
May 21 2016 15:56
I think kmod fixed some of this cases, but I guess nobody really looked into where where the additional refs come from when running the numpy tests. So not sure how hard it is to fix it. But in general I think being more conservative / having more references is not really something I see as a big problem. It's unfortunate that we miss out on some refcnt == 1 optimizations and with numpy there are some user visible problems with it. But still I think it's a smaller issue and something which one has to expect when switching between different python implementations.
I think it's more important to improve the numpy perf a little bit until tuesday so I will look into that :-)
Sun
@Daetalus
May 21 2016 15:57
I see. Thank you!
Marius Wachtler
@undingen
May 21 2016 15:58
do you have any changes which are not yet merged which you want to be included in the release?
Sun
@Daetalus
May 21 2016 15:59
I am try to resolve the last 4 errors. Not include that resize error. When is the dead line?
Marius Wachtler
@undingen
May 21 2016 17:01
cool, I think we want to release it at about tuesday ~6pm UTC. But we need to create the binaries etc so I guess monday would be better as a deadline.
Sun
@Daetalus
May 21 2016 17:02
I see. Thanks!
Marius Wachtler
@undingen
May 21 2016 17:04
oh or it's a problem in BoxIteratorGeneric if this one get's called
Marius Wachtler
@undingen
May 21 2016 17:52
oh wrong chat window...
Kevin Modzelewski
@kmod
May 21 2016 19:09
yep we're planning on cutting the release monday night
I thought we made some progress on the resize / refcnt==1 issue
at least, I mostly-fixed the case that I knew about
if there are more cases or if the fix wasn't complete enough definitely let me know
in general it will be impossible to get 100% exactly the same refcounts as cpython
but if there are cases that we're different I can try to make them more similar
Sun
@Daetalus
May 21 2016 19:42
ok, I create a issue for it @kmod #1196
Kevin Modzelewski
@kmod
May 21 2016 19:47
awesome thanks, I'll take a look
fyi everyone I've read through the master-vs-refcounting diff as much as I can
and we have all the tests passing again!
so unless anyone objects, I'm going to hit merge in a few hours
and Pyston will 100% officially be on refcounting :)
lesshaste
@lesshaste
May 21 2016 19:57
@undingen Cool!
Marius Wachtler
@undingen
May 21 2016 19:58
:-)
:+1: