gEDA-dev: gschem's mkinstalldirs problem

[Stuart Brorson]

>>> /bin/sh @MKINSTALLDIRS@ /home/cnieves/geda/share
>>> /bin/sh: @MKINSTALLDIRS@: File or directory doesn't exist
>> 
>>
>> 	Ack! Another report of this.


> What does "gettext --version" give?  I have 0.14.5.
>
>
> In poking through the generated gschem/aclocal.m4 I see early on in 
> AM_GNU_GETTEXT a call to AM_PO_SUBDIRS and in that macro there is a call to 
> AC_REQUIRE([AM_MKINSTALLDIRS]).  In AM_MKINSTALLDIRS I see:

[snip!]

I don't know the solution to this specific problem, but I can say
this:  On my older FC1 machine I often have problems building gEDA
from CVS since AM_GNU_GETTEXT is not defined anywhere on my M4 search
path.   I'm not an expert, so I don't know why this is.  What I
generally end up doing is locating the relevant .m4 file, and then
setting the env variable ACLOCAL_FLAGS to point to the file's
directory.  [1]

It's possible part of the problem is that AM_GNU_TEXTTEXT is defined
in the /usr/local/ hierarchy, whereas Fedora and RedHat tend to assume
things are located in the /usr hierarchy.

My suggestion therefore is to beef up and make more robust the stuff
in configure.ac which searches out the GNU gettext M4 stuff.

Just a thought . . . .

Stuart

[1]  The problem is that I tend to forget where that directory is
since I rebuild gEDA only every few months on that machine.
Therefore, I need to fool around looking for it each time I update
from CVS.  Maybe I should learn how to take notes . . . . .