[vlc-devel] [vlc-commits] Fix MacOS build

Jean-Baptiste Kempf jb at videolan.org
Wed Jan 25 10:08:23 CET 2012


On Wed, Jan 25, 2012 at 09:08:22AM +0100, Rémi Denis-Courmont wrote :
> This patch modified code paths on five different platforms (Darwin, Linux,
> OS/2, Windows and the rest), including both revectoring and real code
> changes. There is no way that I can test all platforms, especially
> platforms that cannot be cross-compiled to (namely MacOS). I did test it on
> Linux (test suite + smoke test) and Windows. I did retest it after rebasing
> on top of the changes made by others while writing the patch. And it took a
> long time because of the build system churn.

I really think noone blames you on that specific issue.
This is a complex change and hard to test. It got fixed in a timely
manner (in a week), so this is fine.

(to be pedantic, we can CrossCompile to OSX, but it such a pain...)

> I do still make mistakes (e.g. poll), but I do not recall so many

poll was bad. But who actually understand poll, except you (and maybe
Laurent) ?

> complaints about my changes as yours, neither in absolute nor in relative
> numbers.

That is a wrong metric, though... People don't always express
complaints.

> complain the most about your attitude. But I am NOT the only one having

To be honest, we are used to have a quite bad temperature here, but this
has gone too far, lately: we are even worse than FFmpeg vs Libav...
Of course, one of the major blame is me because of the long way of 2.0,
but I cannot do much more than I do. (although I think that if we
shorten our cycles for features, we will shorten testing, but that is
another issue).

I will not start jumping in who is the more annoying, because it is very
pointless and stupid.
However, I think that:
 - we should be more careful about buildsystem changes, because we have
   many people depending on it; seeing the number of people on this very
   mailing list and the number of our users, (over 162million downloaded
   1.1.11). We have this new policy to ask before on the mailing list.
   Shoudl we extend this policy to contrib? or more?
 - we should be very careful on the 2.0 branch as of backporting. I
   think I do it correctly and with a lot of care. So, if you don't want
   to do it, let me do it. The best would be to wait 24hour for backport
   (which is why I do the backports at night).

 - More generally, both of you should calm down.
 - Rafaël, you should excuse yourself for insults on this mailing list.
 - Rémi, you should excuse yourself for insults on IRC.

 - Rafaël, you might want to ask more for feedback when pushing things
   that are sensitive. But I've seen a better trend on that lately.
 - Rémi, you should be a bit more verbose on your remarks and rejection.
   "I already said why this is bad on IRC 9 years ago" is a bit hard to
   get as an answer for a patch :) (yes, exageration here). But I've
   seen a better trend on that lately.


Can we agree to work together a bit more and be more careful about other?
Or else, we do a kernel-development mode...
Or else, I take the project for mine, sell videolan.org for 5 million and
go in the carabbean islands and get free hookers and coke!


   Rémi, your skillz are better than most (all) of us, because you are
   one of the only who _actually_ understand network, threading and unicode.
   I know it is frustrating when you see mistakes that are obvious to
   you, but they are not obvious to the rest of us :)

   Maybe a correct solution is to work on a wiki page to have a
   checklist of the things that are not correct (strtok vs strtok_t,
   etc...) so that people check their patches themselves BEFORE sending
   on vlc-devel for review? I would love to do the page, with you and
   other people, during the FOSDEM.


Best regards,

-- 
Jean-Baptiste Kempf
http://www.jbkempf.com/ - +33 672 704 734
Sent from my Electronic Device



More information about the vlc-devel mailing list