RFC: Error proposal#166
Conversation
|
@samoht @avsm @talex5 @yomimono opinions? others? travis |
|
I cannot comment directly on the diffs as you didn't open a PR, but why did you create a subpackage for pretty-printing? It should just be available in the main package, right? |
|
@samoht lacking knowledge how to get it in mirage-type which doesn't have an mllib file. I opened PR and linked to them above |
|
Yes, should probably be in an mllib file too, which propably depends on topkg-ing the repo first. Thanks for opening the PRs. |
|
I'm dissatisfied with the control flow duplication in mirage-net-unix and mirage-net-solo5... |
|
|
also, point mirage-skeleton back at mirage-dev branch in mirage repo
|
Everything's been merged now. I've updated this PR to remove the pointers to the |
Improve caching on Dockerfile
I started off with
V1.NETWORK(see mirage/mirage#615)result(inlisten,write, andwrtev)errorandstattypes are provided externallyM_pp) shipped asmirage-types.pp) providingpp_network_erroratm (to be extended)TCPIP is the only consumer of
NETWORKand should certainly pass the result up the layer (and do other useful things (such as network interfaces sett.listen <- falseto stop the listen loop), in order to do so, we need to adjust mirage-types further.The error handling in tcpip/tests needs to be adjusted as well (pass it up the layer). I got it up to the point that the tests compile and pass :)
I fixed:
possibly missing some network devices?