and developer guides
Attendees: Amir Chaudhry (chair), Thomas Leonard, Anil Madhavapeddy, Mindy Preston, Dave Scott and Magnus Skjegstad
There's a meta problem where Travis keeps failing and this is making it quite difficult to triage issues. Not had time to deal with this yet.
cohttp - Still want to cut package but there's a backlog of 15 outstanding PRSs on OPAM. Spent some time cleaning up PRs from last few weeks and aiming to get these pushed through in the next few weeks.
mirage-dev - Decision to cut a bunch of releases based on
it's stable and working. That will make it easier to do the next round of
releases but care should be taken to deal with any incoming issues as a result
of these releases. Intent is to stop changing 'v1' at this point. Another
benefit of these releases is that we'll no longer need all the pinned repos to
get ARM working. Incidentally, all known ARM bugs are fixed and ThomasL has
reviewed Dave Scott's patches, so will release miniOS as well.
Fixes are coming in for the TLS stack. We really need to figure out how to get entropy into TLS and there are some thoughts on using vchan to do this.
Also discussed the need for more unit tests, specifically for grant tables etc.
Anil, Dave and Thomas will come up with a plan before end of week as to what to do regarding all the pending releases. Thankfully we've avoided breakages by avoiding partial releases. Question about only supporting 4.02.0 because of conduit. It would benefit a lot from 4.02.0 features and make a number of things much easier.
There was a brief discussion on the mailing list regarding blockers for getting a blog feed on the Mirage website. Essentially, this necessitates date parsing, which someone will need to implement and may be non-trivial.
There's an issue on an external library tracking this if anyone would like to get involved (see Cumulus/Syndic#23).
Removing Xenstore from
mirage-platform. Jon Ludlam still looking at this
and it needs some refactoring. Suspend/resume might also be glitchy.
Xen/ARM hosting: Could be interesting for community to host Unikernels on ARM boards. Could then use that platform to do power measurements and regression tests etc. These could be deployed in people's homes but would require further thought.
Need an Irmin free of
core_kernel. Can then use it for logging layers.
Thomas still working on this.
Next call is scheduled for 14th October - Please refer to the mailing list for actual details a day or so in advance.