Jemalloc 4.0.3 configure script breaks building inside a different git repository.

[Yucong Sun]

I am working around this issue by doing

GIT_DIR=./.git && ./configure

On Wed, Oct 7, 2015 at 3:09 PM, Salvatore Sanfilippo <antirez at gmail.com> wrote:
> Hello,
>
> I was trying to upgrade Redis jemalloc version from 3.6 to 4.0. Redis
> uses its own private copy of jemalloc. Our copy is not modified but we
> have all our dependencies inside /dep so that a given version of Redis
> has a very specific behavior and does not change depending on what the
> user happens to have installed in its system.
>
> So this means that jemalloc, in the case of Redis, happens to be
> inside a directory of a different Git repository (the Redis one).
>
> The new Jemalloc 4.0 configure script replaces the VERSION file
> without doing a sane check that previous versions did, that is:
>
> if test -d "${srcroot}.git" ; then ...
>
> So even if the Jemalloc root directory does not include a git
> repository, but jemalloc just happens to be contained into another git
> repository, the new configure script will attempt to generate a
> VERSION file. As a result it will generate a bogus version file
> matching some commit that happens to contain the specified pattern
> [0-9].[0-9].[0-9].
>
> This in turn means that testing for Jemalloc versions features will break.
>
> In the specific case of Redis I'll just modify the configure script,
> but in general would be good if jemalloc was able to build cleanly
> when embedded into a different Git repository.
>
> Thanks for all the work that went into the 4.0 release, we hope Redis
> users will benefit from this ASAP. Btw for us the ability to configure
> size classes from the configure script was a good change.
>
> Cheers,
> Salvatore
>
> --
> Salvatore 'antirez' Sanfilippo
> open source developer - Redis Labs https://redislabs.com
>
> "If a system is to have conceptual integrity, someone must control the
> concepts."
>        — Fred Brooks, "The Mythical Man-Month", 1975.
> _______________________________________________
> jemalloc-discuss mailing list
> jemalloc-discuss at canonware.com
> http://www.canonware.com/mailman/listinfo/jemalloc-discuss