[Slackbuilds-users] Source code missing for some Slackbuilds scripts

[B. Watson]

On Mon, 10 Mar 2025, Luveh Keraph wrote:

> Yesterday I was building a bunch of packages in a system to match what I have in a separate system. As it happens, a substantial number of them - like e.g. calc, makemkv, monitorix,
> transcode, pari) could not be built simply because the source code to be used is not available in the target location any more. 

There's a solution for this: the SBo source archive.

https://slackware.uk/sbosrcarch/

slackbuilds.org itself doesn't host sources, so the archive is hosted
elsewhere. Every time there's an update to the SBo master git branch,
the new DOWNLOAD and DOWNLOAD_x86_64 files are downloaded and filed in
the archive trees (by-name and by-md5).

If you use sbopkg, it will automatically use the archive if the
original URL doesn't work. Also, if you use sbo-maintainer-tools, it
has a sbodl command that can download sources from the archive (the
-a option).

> For the most part it is not a big deal, for the new sources can usually be found without a problem (things for transcode were a bit more tricky) and a simple modification to the
> corresponding Slackbuilds script gets things built properly. I was wondering, however, whether it would be a good idea to add small script to the Slackbuilds repository whereby the
> script attempts to download the source code for each item in Slackbuilds once a day every day, and sends a warning email to the maintainer of each package.

If someone implements this, I beg of you, keep track of previous
emails and don't send the maintainer the same email twice. Also, daily
is a bad idea. Weekly would be better.

This is because the repo only updates weekly, on Friday nights. So if
a source disappears on Saturday and I get an email about it, I fix the
build and submit the update... but it won't get merged to the master
branch until the next Friday. I do *not* want to receive 5 more emails
that week telling me to update my build, since I already did. If your
system did that, then I probably would block the emails or file them
as spam, for my own sanity.

Another issue with your idea... some of the sources are rather
large. Constantly re-downloading them every day (or even every week)
might cost us money.

Hard numbers: The sbosrcarch by-name and by-md5 directories are each
190GB. Files are hardlinked between the two trees, and outdated files
are removed from the archive weekly, so this gives a pretty good idea
of how much source your email script would have to download per run.

190GB a day would be 5.7 terabytes a month...

Of course your script could just do an HTTP HEAD instead of a
download. But there are web servers out there that refuse to respond
to a HEAD request. Plus, it doesn't let the script detect when the
md5sum is wrong (meaning, the file changed on the server).