Re: guppi problem

[Havoc Pennington <rhp zirx pair com>]

On Sun, 13 Jun 1999, Thomas Shannon wrote:
> As a biochemist, I have an intense interest in this program.  And yet I
> have never been able to get it to compile.  I have been trying since the
> first release.  Nevertheless, I got so close this time I thought I would
> post and see if someone could help.
>

Please, always post as soon as you have a problem! :-) If no one tells me
that there are problems, then I can't fix them. If you post, you'll help
everyone else out too.

Do keep in mind that Guppi is still in development, so you probably can't
(quite) use it for biochemistry yet. :-) But I certainly hope you will be
able to soon, and I'm interested in what features you need.

I should also write up a little bit of text about how to use the
functionality that exists, since it can be hard to know what's intended.

> The primary failure this time seems to be related to the absence of
> something called "guile-snarf" files.  These apparently have the file
> extension ".x" If they are supposed to be with the source files, I can't
> find them.
> 
> Anybody know what's going on?
>

These files are supposed to be generated at compile time. I've had some
problems with this, but I thought I had fixed it in more recent releases.

I don't know if you want painful details about the problem I know about,
but basically when automake creates the .tar.gz, it's supposed to generate
a Makefile that knows that foo.cc depends on foo.x (because foo.cc
#includes foo.x). As I understand it, it does this by slurping the ".deps"
files found in a maintainer build tree and hard-coding them in the
distributed Makefile. automake does this when you pass it the
--include-deps option. However, automake has been refusing to do this; it
appears to simply ignore --include-deps. So the Makefile fails to generate
the .x files when you type make.

I thought I'd worked around this by including the .x files in the .tar.gz
distribution, so you don't have to generate them. However, if you 'make
clean,' you will probably blow away the .x files and the Makefile won't
know how to regenerate them. Goose has the same hack in it, so I'm happy
to hear Goose compiled (hopefully Goose compiled with Guile support - you
wouldn't have encountered the problem if Goose's configure script didn't
find Guile).

Let me know if this sounds like a likely cause of your problem, and the
exact error message you get if not. I should have put a warning about this
in the Guppi announcement; I'll do so in the future.

Havoc