Looks good to me.

Check last_update date extraction for a package.

LGTM

LGTM

Looks good to me, thanks !

Based on my tests on our Jenkins instance, swh-web build jobs with the changes in that diff become unstable when there is more
than three jobs executed in parallel (as there is a lot of cypress processes consuming a lot of resources on the Jenkins node).

In T3423#67264, @anlambert wrote:

After some hacks on the current Jenkins pipeline for swh-web to test if it was feasible, I managed to setup the use of sorry-cypress but could not run multiple cypress in parallel as Jenkins pipelines forbid nested parallel stages.
So this would require setting up a dedicated pipeline build job for cypress tests as currently Python and cypress tests are executed in parallel in a single job.

For the record, that new doc is not build by our CI, apply this patch to do it.

diff --git a/tox.ini b/tox.ini
index 3e768bf..52f8678 100644
--- a/tox.ini
+++ b/tox.ini
@@ -16,6 +16,7 @@ setenv =
 commands =
   {envpython} -m pifpaf run postgresql -- make -C sysadm {posargs:html}
   {envpython} -m pifpaf run postgresql -- make -C user {posargs:html}
+  {envpython} -m pifpaf run postgresql -- make -C general {posargs:html}
   {envpython} -m pifpaf run postgresql -- make -C docs {posargs:html}

Why duplicating loader/lister logos ? I do not get its purpose.

Fix docstring

I found another issue when trying to run the loader in our docker env, see inline comment.

I found an issue when testing the lister in our docker environment, see inline comment.

Please add an incremental parameter to the lister in order to be able to perform a full relisting of go packages if it is set to False
(other listers with incremental feature proceed like this).

Rebase

No significant speedup when running tests, abandoning this.

Not really conclusive, trying with /dev/shm/.

Add comments and remove timeout config

Yrying sqlite WAL journal approach.

In D8300#215886, @vlorentz wrote:

Actually, what about changing the DB path, so both tests instances don't use the same? :memory: might even work

Looks good to me, thanks !

Looks good to me.