Re: [Denemo-devel] The missing regex in mxe build

[Richard Shann]

On Tue, 2013-02-26 at 09:50 -0600, Jeremiah Benham wrote:
> I was using the release 0.9.6 or whatever it is called with the
> denemo.py file I sent and the rc10 tarball. The mingw branch needs
> some correction. that's on my todo list.

Hmm, I see I have caused some confusion here by naming branch mingw
locally, that is I did:

git checkout -b mingw origin/release-mingw-0.9.6

so the branch I am using *is* the release-mingw-0.9.6 branch which in my
local repository is called just mingw. I have just checked that this is
really the case by looking in my denemo.py and checking that it has the
release-mingw-0.9.6 lines that comment out jack and lash, while your git
mingw branch does not have these lines at all.

So, I am using the same branch as you, with the only change being to
delete the ABI=32 which you had inserted with
b942d5bb2cf1dcd3812c548a488655def4edf501 seven months ago, which stops
my 64 bit system from building gmp.

The python failure is strange because looking back through the log there
is nothing to indicate that anything went wrong before it tried to
install.

Richard


> 
> Jeremiah
> 
> On Feb 26, 2013 8:49 AM, "Richard Shann" <address@hidden>
> wrote:
>         On Fri, 2013-02-22 at 15:07 -0600, Jeremiah Benham wrote:
>         > That problem can surely be fixed.
>         I traced the problem with gmp back to the ABI=32 which had
>         been added to
>         specs/gmp.py
>         
>         Fixing this, and the intl* binary finding problem I got to
>         this:
>         
>         
>         python: error while loading shared libraries:
>         libpython2.4.so.1.0:
>         cannot open shared object file: No such file or directory
>         
>         during the install of tools::python
>         
>         Any ideas on this one?
>         
>         (to recap, this is make denemo on the gub from mingw branch of
>         jjbenham's githost repository).
>         
>         I am hoping that if I can build denemo for windows I will be
>         able to
>         move gdk/gtk/ on to a version that works on the various
>         windows boxes I
>         can test on. (I am reasonably confident that the crashes on xp
>         boxes are
>         due to some bug in gtk).
>         
>         Richard
>         
>         
>         > You might be able to use master branch gmp.py or something.
>         I gub up a
>         > new snapshot. Its called rc10. Get it at the usual place.
>         >
>         > Jeremiah
>         >
>         > On Feb 22, 2013 1:12 PM, "Richard Shann"
>         <address@hidden>
>         > wrote:
>         >         On Fri, 2013-02-22 at 17:33 +0000, Richard Shann
>         wrote:
>         >         > I'll try and build myself from your gub using the
>         branch on
>         >         your gub
>         >         > branch named "release-mingw-0.9.6".
>         >
>         >         Very curious - this fails in gmp with this in the
>         log:
>         >
>         >         checking ABI=32
>         >         checking compiler gcc
>         >         -I/home/rshann/gub/target/tools/root/usr/include
>         >         -D_GNU_SOURCE... yes
>         >         checking compiler gcc
>         >         -I/home/rshann/gub/target/tools/root/usr/include
>         >         has sizeof(long)==4... no
>         >         checking compiler icc
>         >         -I/home/rshann/gub/target/tools/root/usr/include
>         >         -D_GNU_SOURCE... no
>         >         configure: error: could not find a working compiler,
>         see
>         >         config.log for
>         >         details
>         >
>         >         So on this 64-bit system it seems unable to cope.
>         What is
>         >         rather bizarre
>         >         is this is the first time I've encountered this
>         despite
>         >         building gmp
>         >         several times (most recently using the master branch
>         of the
>         >         gub).
>         >
>         >         Well, if you at least can build it that may be
>         enough.
>         >
>         >         Richard
>         >
>         >
>         
>