Re: [O] HTML export fails with (void-variable org-version)

[Achim Gratz]

Bernt Hansen writes:
> I tried that... and installed emacs under Cygwin (so it can find emacs
> to run emacs -batch) and then had to fix permissions on org-install
> (since it's on a windows drive) before make autoloads worked.

It's preferrable to use NTEmacs for this step, even though the .elc
files should be compatible.

> I didn't spend any time trying to correct install locations in the
> makefile.  I run org-mode without compiling any sources normally and
> having to do the extra make step each time I move around the source tree
> isn't going to be very enjoyable.  I have no idea if I need to rerun
> this make autoloads command on a regular basis or not.

Yes and no.  As long as all the autoloaded functions continue to be in
the same load-file and their names don't change and there are no new
functions that need to be autoloaded, org-install.el effectively isn't
going to change.  But that's a deception, since there is no way you
could determine if it should have been changed unless you actually
generate it anew.

As I said, I'm still working to provide specifically for the case that
org-version.el is missing (which would normally produce an autoloaded
defconst org-version...) and injecting an org-version from looking at
the git version of the work-tree (if it finds one).  That will also
encompass an (autoloaded) function (with the load-file located in
UTILITIES/ most likely) that you can call in emacs to re-generate
org-install.el, since org-install.el is absolutely not optional.

So going forward make will not be a requirement for simply using org
either from Git or from tarball anymore.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Wavetables for the Terratec KOMPLEXER:
http://Synth.Stromeko.net/Downloads.html#KomplexerWaves