Inspired by Producing OSS, I'm taking this back to the list.
Gavin 'Beau' Baumanis <gavinb_at_thespidernet.com> писал в своём письме Sat,
12 Sep 2009 04:35:44 +0400:
> Hi Roman,
>
> Do you have commit rights for that area of the repository?
> If you have, then you've ben afforded those privileges because you're
> trusted to do things correctly.
>
> (I understand your desire to have some extra eyes look over it).
> But not withstanding that... since no one else has yet managed to get to
> it and as long as
> you've built it against trunk etc...all the normal things... then Surely
> it is appropriate to commit it?
>
> if you're still a little hesitant to commit, then at the very worst, you
> could create a ticket and attach the code to it and then reference your
> ticket number back here in this thread.
>
> But again - if you have commit rights... then (especially now that it
> has been two weeks) simply commit it.
> If it causes an unforeseen issue down the track - it can always be
> reverted.
>
>
> Gavin.
Gavin,
That depends on the meaning of "commit rights". I can commit all over the
repository, but the build system is outside my delegated zone of authority.
HACKING doesn't seem to imply that no approval allows a partial committer
to commit outside of his sandbox, and seeing as the build system is an
important part of the project, I'd rather not assume that.
Meanwhile, I noticed that the Subversion libintl is specifically modified
not to link to libiconv, which means it's not going to recode the strings
whether bind_textdomain_codeset is called or not. Huh. I still suppose an
explicit call won't hurt.
Roman.
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2394136
Received on 2009-09-13 01:04:47 CEST