Re: [RFC-v2] First scripts for ARI support inisde gdb tree

[Joel Brobecker <brobecker at adacore dot com>]

> It's called from within update-wevb-ari.sh,
> I would like to leave that script
> as close as possible to the one in ss directory 
> on sourceware, so that we can just call that script if it exists in the cvs 
> checkout made to create the daily snapshot.
[...]
> I don't want to remove it completely, 
> as I would like that the nightly snapshot script still generates
> those messages when run.

I think that this is where the confusion is. From my perspective,
what I would like in the GDB sources is a script, or a set of scripts
if we have to, that generates the ARI data in a very simple way.
that's why I said, just generate it in current directory, or a
subdirectory if we'd like.  That way, a user wanting to generate
the ARI would just do:

        % /path/to/gdb/contrib/ari/generate-ari.sh
        % web-browser ari/index.html

... and voila!

Once we have that, we (I) will modify the scripts in ss to call
that new script and then put a copy of the generated HTML at
the right location. Similarly, it will do the diff and send an
email if new regressions have appeared.

I do not mind it if we get to this stage in baby steps. In other
words, if you start by having something close to what what in ss,
and then slowly modify it to reach the goals above. We do not
have to use the new gdb/ari scripts from ss immediately; we could
keep using the ones in ss until we feel the ones in gdb are ready...

You can take the above as suggestions, not requirements for approval.
In the end, you will be the one responsible for these scripts, so
I'll let you decide how you'll want to design all this.

-- 
Joel