Re: Separate obj tree

[Doug Alcorn]

Gary Oberbrunner <address@hidden> writes:

> However, if you're using emacs (on Unix, with links) you have to have
> things set up just right otherwise you can get two buffers on the same
> file (via the different link names).  

Are you saying that you have figured out how to overcome this problem?
Please share.

> I do understand the #include-problem with generated headers (which I
> have in abundance), so I haven't tried to change the way cons works.
> Ideally I'd write an emacs function that would process the error
> filename in a configurable way before jumping to it.  If only I had
> time...

I looked into this for a day or so.  I could never find the right hook
to add a function to.  My strategy was simply parse the compilation
buffer and delete occurances of the build directory.  So, on my system
I have lines like this:

TTServer/Linux/build/commmon/CbDbTblServerDb.cpp: Warning....

and I was trying to get:

common/CbDbTblServerDb.cpp: Warning...

This is obviously not very generic.  My setup just happens to allow
for it.  But, like I said, I couldn't find the right hook to insert
this function.  The problem was that the compilation mode had already
parsed the buffer and stored off where the errors/warnings were
located.  Changing the buffer at that point didn't do any good.  What
I really need to do is make my changes to the compilation buffer
before it is parsed for errors/warnings.
-- 
 (__)  Doug Alcorn (mailto:address@hidden http://www.lathi.net)
 oo /  Unix Hacker
 |_/   "It's too late for paradise"