This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Remove PRI_MACROS_BROKEN define usage
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: Adhemerval Zanella <azanella at linux dot vnet dot ibm dot com>, libc-alpha at sourceware dot org
- Date: Fri, 02 May 2014 13:53:03 -0400
- Subject: Re: [PATCH] Remove PRI_MACROS_BROKEN define usage
- Authentication-results: sourceware.org; auth=none
- References: <532B3168 dot 2030108 at linux dot vnet dot ibm dot com> <Pine dot LNX dot 4 dot 64 dot 1403201832040 dot 29670 at digraph dot polyomino dot org dot uk> <532B3750 dot 2030102 at linux dot vnet dot ibm dot com> <Pine dot LNX dot 4 dot 64 dot 1403201848000 dot 29670 at digraph dot polyomino dot org dot uk> <532B3F64 dot 6030005 at linux dot vnet dot ibm dot com> <53634D52 dot 8010502 at redhat dot com> <Pine dot LNX dot 4 dot 64 dot 1405021511400 dot 1748 at digraph dot polyomino dot org dot uk>
On 05/02/2014 11:12 AM, Joseph S. Myers wrote:
> On Fri, 2 May 2014, Carlos O'Donell wrote:
>
>> Even though Steve Ellcey has gotten upstream to change the code
>> such that PRI_MACROS_BROKEN is always defined, there is no stable
>> release with that change yet.
>
> I don't think gettext releases are relevant to us; it's the development
> sources of glibc's libintl that we should aim to have in sync with the
> development sources of gettext's libintl, by merging local changes in both
> directions and implementing things in such a way that identical code can
> work in both places.
May I suggest the following plan of action then?
* Merge libintl 0.18.3.2 (official release) into glibc and fixup
where appropriate, testing the result.
* Pro: You know libitinl works because it's a released version.
* Con: Subsequent merge from glibc to gettext is harder.
* Review remaining differences between glibc master and gettext
master and propose patches.
This reduces the merge-in risk of breaking glibc by using
a stable gettext, but allows us to iterate down to zero
differences.
The 0.18.3.2 differences aren't large, but they still have
new generic lock thread support, and pathname support, and
merging this in worries me slightly. That worry is why I
recommend a conservative first step.
Are you suggesting that this is really just a waste of time
given libintl's stability and doing the above is just 2x
the work for no reward?
Cheers,
Carlos.