[PATCH rsa/stdint_noheaders] Cleanup: Add #include <stdint.h> for uint[32|64]_t usage (except installed headers)

[Thomas Schwinge]

Hi!

On Tue, 04 Jun 2013 11:47:23 -0400, "Carlos O'Donell" <carlos@redhat.com> wrote:
> On 06/04/2013 11:44 AM, Ryan S. Arnold wrote:
> > On Thu, May 30, 2013 at 12:23 PM, Ryan S. Arnold <ryan.arnold@gmail.com> wrote:
> >> On Sun, May 26, 2013 at 2:40 PM, Thomas Schwinge
> >> <thomas@codesourcery.com> wrote:
> >>>>>> On Wed, May 01, 2013 at 11:23:39AM -0500, Ryan Arnold wrote:
> >>>>>>> Per Roland's suggestion in "[PATCH rsa/stdint] Cleanup: Add #include
> >>>>>>> <stdint.h> for uint[32|64]_t usage" I have removed the installed headers
> >>>>>>> from this patch, and separated the ChangeLog entries into the proper
> >>>>>>> subdirectory ChangeLogs.
> >>>
> >>>> Pushed upstream:
> >>>>
> >>>> commit e054f494306530720114b321b3d97ca2f397cbbb
> >>>> Author: Ryan S. Arnold <rsa@linux.vnet.ibm.com>
> >>>> Date:   Wed May 1 10:46:34 2013 -0500
> >>>>
> >>>>     Add #include <stdint.h> for uint[32|64]_t usage (except installed headers).
> >>>
> >>> I'm seeing:
> >>>
> >>>     make[2]: Entering directory `[...]/locale'
> >>>     /usr/bin/perl gen-translit.pl < C-translit.h.in > C-translit.h.tmp
> >>>     mv -f C-translit.h.tmp C-translit.h
> >>>
> >>> ..., resulting in: [...]

> >>> Should a) the change be reverted for this file (that is, the above hunk
> >>> be applied), or b) the generator be changed to emit that #include, or,
> >>> heck, c) the autogenerated file be removed from the source repository
> >>> (and then some)?

b) has been done.

> > I suppose Thomas's question still stands.  Should the autogenerated
> > file be removed from the source repository?

Yes, that question still stands.

> We shouldn't need to force everyone to regenerate
> files on their target system, it would really complicate the build
> process.
> 
> Feel free to argue differently though.

Well, as you can see from my »I'm seeing« snippet above, the build
process already happily does recreate the generated file (when its
Makefile dependencies tell to do so, I suppose), so -- assuming the
generator reliably recreates the file and does not have any special
run-time requirements (which might be the argument for keeping configure
files checked in) -- there is no point in keeping the generated file
checked in other than saving some really minimal amount of build time.

Is this worth a policy change (and subsequent action, for all generated
files that are kept checked in; the list of which is still to be
determined)?


Grüße,
 Thomas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-alpha/attachments/20130620/212d4708/attachment.sig>