xmo-odoo

Also needs a reverse index endpoint, so e.g. the runbot can pass in a commit (or a set of commits, most likely), and get the batched PRs from that staging....

On the back burner after implementing (and soon deploying) the reverse index bits for the runbot: generating a git repo / pack on the fly works but it's *super slow*...

New update: a basic wip script seems to process 84589 stagings in ~35 minutes. That is lacking *batches* support, and eyeballing it there seems to be an average of around...

Notes from initial (completely failed) attempt: - some of the crons clearly have non-obvious dependencies, so the cron triggers should probably be evaluated and implemented individually to not get completely...

Having thought about / fiddled with this a bit more, I think I've gathered why the previous effort died miserably as I did a straightforward conversion then got left with...

Pulling back because I want to do it, but - it's hardly urgent (more of a nice to have to compensate for a few gh deficiencies) - it requires a...

> In comparison, how long does it takes for the current ones? 1min processing could quickly be an issue with 1000 devs pushing all over the place. OTOH that's to...

Also if the first split succeeds and the second fails, we can probably re-split the failed one immediately without restaging it in full, so our wallclock gain is probably closer...

> So not sure of the added value Well I guess there's the value that you *could* craft the initial such that all forward ports succeed rather than discover the...

There's a new request for this from the runbot team, in order to better handle exceptions and batched exceptions: currently it's difficult to know whether a non-trivial commit to stable...