Re: Spam message when using CVS for webpages

[Ian Kelling]

Ineiev <ineiev@gnu.org> writes:

> The problem is, we don't deploy the exactly same version for all
> Savannah hosts at once, we update them one by one, so you hardly
> would be able to tell which Git commit corresponds to software
> running on the particular host; this feature makes sure the users
> can download the right version.

I definitely admire the ingenuity to offer source code in more
places. However, I'm pretty confident Savannah webpages are a sufficient
place to satisfy the AGPL requirement of offering source, and adding
output like this to command line operations where the only expected
output is information related to the operation is undesirable for
various reasons and will very likely cause breakage for scripts and
tools which make calls to Savannah.

For the problem of different machines having different source, the link
for source at the bottom of savannah webpages could say something like:

"Savannah source repository is here: http://XXXX. Savannah is split onto
several machines, and the code running on some machines can lag behind
what is in our repository. Here is how to get the exact versions being
run:

To get the source code on the machine handling cvs requests, run

rsync -avz --cvs-exclude gavin@cvs.savannah.nongnu.org:/opt/src/savane .

To get the source code on the machine doing X, run ... (fill in more
here)"

Especially because this is likely to break other tools and annoy people,
I think it should be reverted until there is some consensus among
savannah hackers on the right solution.