Consistent autoconf version

[Chris Hegarty]

On 26 Aug 2013, at 02:56, David Holmes <david.holmes at oracle.com> wrote:

> On 25/08/2013 6:34 AM, Mike Duigou wrote:
>> 
>> On Aug 24 2013, at 05:52 , Chris Hegarty wrote:
>> 
>>> Hi,
>>> 
>>> Sorry if this has come up before, but I cannot find anything in the archives.
>>> 
>>> Is it possible to agree an autoconf version to use to generate the checked in generated-configure.sh?
>>> 
>>> I recently had to merge, and regenerate this file, and found versions that were generated with 2.67, 2.68, and 2.69. Using such different versions makes it nearly impossible to see the relevant diffs in generated-configure.sh. While using the same version greatly, but not completely, simplifies the number of changes.
>>> 
>>> Is there any reason why not to agree a specific version?
>> 
>> My understanding was it is difficult to use a non-standard version on some (most?) platforms. Most people would probably balk at having to use any version other than that installed by default on their system. Alternatively, if we want a specific version we could mandate that the checked in generated-configure.sh be pushed by a project owner with access to the "official" version. To my knowledge there's only been one buggy (due to buggy autoconf which generated it) generated-configure.sh been checked in. The diffs are annoying but in theory you're not really supposed to read generated-configure.sh, instead you should be reviewing the autoconf inputs which generate it.
> 
> Right - you don't look at the generated files. :) So you never attempt a

Fingers crossed!

> filemerge you simply regenerate from the current "sources". If hg wants to merge the generated file let it do its thing then:
> 
> hg revert -C generated-configure.sh
> (cd common/autoconf && sh autogen.sh)
> hg resolve -m generated-configure.sh
> hg commit -M

Thanks. Yes I get this, just wondering why we don't at least "try" to stick to a specific version. It would make skimming the diffs much easier. As well as reduce the noise in mercurial history.

-Chris.

> 
> David
> -----
> 
>> YMMV,
>> 
>> Mike
>>