[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Libvir] [PATCH] Make "make distcheck" work.



Daniel Veillard <veillard redhat com> wrote:
> On Wed, Nov 14, 2007 at 03:57:10PM +0100, Jim Meyering wrote:
>> Here's another infrastructure-fixing patch.
>> I got most of the way to an honest "make distcheck",
>> but fixing a final failing sub-test wasn't worth the effort,
>> so I punted and added the top level "distuninstallcheck"
>> target to make it skip that part.

Hi Daniel,

Thanks for the thorough review!

>> 	Make "make distcheck" work.
>>
>> 	 * Makefile.am: Expand some "*" wildcards, and (for now) disable
>> 	   the relatively unimportant, distuninstallcheck target.
>> 	   Fix a few redirect-directly-to-target bugs.
>> 	   Add a few $(srcdir)/ prefixes and add an uninstall-local rule.
>> 	 * docs/Makefile.am: More of the same.  Split some long lines.
>> 	 * docs/examples/Makefile.am: Likewise.
>> 	 * python/Makefile.am: Likewise.
>> 	 * python/tests/Makefile.am: Likewise.
>> 	 * qemud/Makefile.am: Likewise.
>> 	 * tests/Makefile.am: Remove the directories already listed in SUBDIRS.
>
>   okay, lot of this makes sense. The only drawback seems to be loosing
> the ability to use wildcards to list for example the HTML pages or
> C examples ...

At first, it does seem like using wildcards is easier than listing
the files explicitly, but listing them has advantages:

  - with wildcards, one can accidentally distribute/install files just
    because there was a spare .c, .xml, .html file in the wrong directory.
    With explicit lists, that can't happen.

  - with the explicit list, you have a record in version control of
    which files were distributed: less reliance on state of build
    directory, hence more reproducible.

Note that if you forget to add the name of a new file to a list,
there is little chance it'll make it past "make distcheck", as long
as something depends on the new file.

>   BTW, have you noticed that docs/examples/Makefile.am is automatically
> generated, so for that part you will instead need to fix index.py which
> does the indexing/generation.

Oh!  Thanks.  I didn't know that.  Will adjust.
At the same time, I'll tweak the build rules so that
the generated file is read-only.  That should serve as a warning
to anyone else who is tempted to modify that file directly.

>    okay in principle but the docs/examples/Makefile.am change need to be
> fixed differently.

New patch coming up...


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]