Re: [Tinycc-devel] Define __REDIRECT_NTH

[Kirill Smelkov]

On Mon, Sep 13, 2010 at 05:46:02PM +0200, grischka wrote:
> Kirill Smelkov wrote:
> >On Sun, Sep 12, 2010 at 01:17:07AM +0200, Thomas Preud'homme wrote:
> >>Greeting everybody,
> >>
> >>I don't know if I should announce every commit I push to mob but here is 
> >>another commit.
> >
> >[...]
> >
> >Let's maybe add something to "Commit notify - mail to" field at
> >http://repo.or.cz/editproj.cgi?name=tinycc.git ? (*)
> >
> >I propose we just add address@hidden as destination for commit
> >notification mails - this way it would be handy for everyone for keeping
> >an eye about changes, and also it would be possible to directly discuss
> >proposed-to-mob commits on the list.
> 
> I've only recently thought about activating commit notification.  I don't
> have a strong opinion on this, but actually what I'd like to have sometimes
> is more background information and not so much the technical detail which
> anyway is already present in the commit itself.
> 
> For example I'm often curious:  What were people doing when they discovered
> bug X or missing feature Y.  What package did they try out on tinycc, what
> platform are they working on.  Or with some not quite common feature: why
> do we need that, what is it good for.  All such things.  Of course I could
> google, but I'm lazy.

I guess what you describe should be a good rule on how to write log
messages, at least briefly.

For example I wanted tcc to be faster gcc, and that's there

http://repo.or.cz/w/tinycc.git/commit/9775aea1ab634a74cd4254df85e93d4cbf098cce

though I've ommitted context in

http://repo.or.cz/w/tinycc.git/commit/7901d1e3ad19b630c39388612a5ad7c0bf84df39

my bad... - will try to improve next time when I have a chance :)

> In that sense I think some personal notes are a good thing, hence and when.
> Not that every single commit needs to be commented on the mailing list.
> Nobody wants to be spammed after all ;)  YMMV ...

They sometimes also create secondary -commits mailing list, and email
notification goes there, and when one wants to discuss some commit,
he/she replies to both -commits and main (e.g. tinycc-devel in our case)
list.

This way main list is not "spammed", but when someone wants to
comment/announce he has everything at hand.

YMMV of course (I personally study git log of projects of medium
interest on thursdays in batches ...)

> 
> >Other options exist too, of course...
> >
> >Thanks,
> >Kirill
> >
> >
> >(*) Grischka, I forgot admin password (and now I'm not the admin, or
> >    else I'd directly try it myself :) ), so I cc you too.
> 
> (Password sent off list)

Thanks :)

> As to notifications from repo.or.cz, basically what you would get is this:
> 
> >This is an automated email from the git hooks/post-receive script. It was
> >generated because a ref change was pushed to the repository containing
> >the project xyzzz.git.
> >
> >The branch, mob has been updated
> >       via  40a8cf3c4192cb6862de14fe8a393324aa655f86 (commit)
> >      from  ed0fb7ea3538bde77f222b83c6869913e8dc8a59 (commit)
> >
> >Those revisions listed above that are new to this repository have
> >not appeared on any other notification email; so we list those
> >revisions in full, below.
> >
> >- Log -----------------------------------------------------------------
> 
> ... commit log (including diffs) follows.

I guess I know how to tweak it to produce something like below example
for each patch (yes, that's spamming in your speak :))

---- 8< ----
Subject: [tinycc.git:mob] Add support for __REDIRECT_NTH needed with eglibc.
From: commit-bot for Thomas Preud'homme <address@hidden>
To: address@hidden
Cc: Thomas Preud'homme <address@hidden>

Commit-ID:  2596273fce52e1f9db1b03d495537614fd56cef5
Author:     Thomas Preud'homme <address@hidden>
AuthorDate: Sat, 11 Sep 2010 17:57:15 +0200
Commit:     Thomas Preud'homme <address@hidden>
CommitDate: Sun, 12 Sep 2010 01:19:19 +0200

Add support for __REDIRECT_NTH needed with eglibc.

Add support for __REDIRECT_NTH as eglibc makes use of this macro to
redirect long double functions to long functions on arch not supporting
long double.

---
 libtcc.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/libtcc.c b/libtcc.c
index a89ecc3..a755708 100644
--- a/libtcc.c
+++ b/libtcc.c
@@ -970,6 +970,9 @@ LIBTCCAPI TCCState *tcc_new(void)
 #else
     tcc_define_symbol(s, "__WCHAR_TYPE__", "int");
 #endif
+
+    /* glibc defines */
+    tcc_define_symbol(s, "__REDIRECT_NTH(name, proto, alias)", "name proto 
__asm__ (#alias) __THROW");
     
 #ifndef TCC_TARGET_PE
     /* default library paths */
-- 
1.7.3.rc0.31.g5b4a4

---- 8< ----