This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: [PING^2] Re: [PING/tiny respin] Re: [PATCH, 0...4/4] Enable libstdc++ as a DLL on windows platforms.
- From: Dave Korn <dave dot korn dot cygwin at googlemail dot com>
- To: Jonathan Wakely <jwakely dot gcc at gmail dot com>
- Cc: Dave Korn <dave dot korn dot cygwin at googlemail dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>, libstdc++ at gcc dot gnu dot org
- Date: Mon, 10 Aug 2009 21:28:16 +0100
- Subject: Re: [PING^2] Re: [PING/tiny respin] Re: [PATCH, 0...4/4] Enable libstdc++ as a DLL on windows platforms.
- References: <4A60F7BA.3040204@gmail.com> <4A6FCDB1.4050504@gmail.com> <4A7F29A7.9010508@gmail.com> <4348dea50908101224x24f384d6r68dc3acf12048674@mail.gmail.com>
Jonathan Wakely wrote:
> 2009/8/9 Dave Korn:
>> Ping; part of this patch is still awaiting review by a v3 maintainer, and the
>> clock is starting to tick on stage 1, so I've remembered to Cc the v3 list this
>> time.
>
> I can't approve or reject,
Your comments are still welcome :)
> but have some questions. As a v3
> developer, how do I avoid breaking cygwin in future, without having to
> learn how DLLs work? e.g. I would have guessed the manipulators
> ends/endl in include/bits/ostream.tcc would also be marked with
> _GLIBCXX_IMPORT, but they're not. If I add new code, can I ignore
> _GLIBCXX_IMPORT and rely on a cygwin maintainer to fix any breakage?
Yep, you can; omitting an instance is non-critical, but makes for less
efficient code that relies on ld's auto-import facility, rather than being
generated by the compiler with the knowledge that it will be imported at link
time in the first place. ld's auto-import facility works well enough, but is
mildly inefficient and in the long-term we'd like to turn it off by default,
which will work just fine once we've got all the right markup in place.
ends/endl may well be an omission on my part, I'll check into that later and
send a follow-up patch if necessary.
> Looks like you've made some lines >80 characters, that goes against
> the C++STYLE standard.
Yeh, there were a couple that were a few chars over. I thought it'd be easier
to ask forgiveness... they're only a little over, and the rest of the
declarations around them are all one-per-line, so I figured I'd ask if anyone
minded.
cheers,
DaveK