[Bot Message]
One or more PR numbers were detected in the summary.
Links to those PRs have been inserted into the summary for reference.

In D7071#170268, @Fabien wrote:

Note that you still need to build an index per group (so within the loop), despite this is not what is done here.

This doesn't strike me as something that belongs in CONTRIBUTING and brings up lots of questions:

• Why do we expect new devs to have multiple shellcheck versions installed?
• Why do we expect shellcheck to not be in PATH correctly?
• What makes shellcheck install special such that we need to tell devs to check the version, but we do not do this for other tools?

Expanding our ownership to gitian-builder makes me uncomfortable, but this change is small and the benefits far outweigh the cost of maintaining this change.

Version bump is missing from contrib/debian/control. Please add this to the test plan also.

Use a JOB_POOL mechanism to limit the number of concurrent jobs.

I don't see the value of this diff. IMO a shorter URL (3 chars !) is not a strong case and this:

• Makes the url inconsistent with the file system scheme,
• Will make it more complicated to add translations, if someone is willing to do so.

Note that you still need to build an index per group (so within the loop), despite this is not what is done here.

The path is finally what Teamcity expects. You can see the result here.

Use version from the artifact dir.

Make the path relative to the checkout directory

Try to send the message before copying the result file, as Teamcity doc says it monitors for the files to appear

Duplicated junit/ in path

Comment about the new approximation for exponential:

Update importdata message.
Since the doc contradicts itself some more iterations are expected...

In D7057#170064, @Fabien wrote:

I don't quite get it, since D7052 updated the generated extension there is no html anymore in the RPC docs ?

There is a clearly some previous backports that are missing.

Update comments

Simplify target computation by dropping one bit out of the exp2 compuation. That bit is not significant anyways, so we are good.