<Drup>
hannes: the second reason I like your patch a lot (on top of being really nice) is that I'm just happy people start hacking on functoria's insides :D
<hannes>
Drup: I still don't understand much of functoria
<hannes>
but found some aggressive energy and used it to solve some problem which was lingering for way too long
<hannes>
mato: any chance to get the LD stuff in soon (will save myself rebases, but of course I appreciate another pair of eyes)?
<hannes>
Drup: and yes, the leftover record... that was me not cleaning up enough stuff, but it was already 5 in the morning and then I forgot about it
<hannes>
Drup: there's also a slight chance to get the version comparison from opam relicensed under ISC and just include that in functoria :)
<mato>
hannes: ah, LD, yes, sorry. Was going to get to that but got distracted trying to review and understand mirage#691
<hannes>
mato: yeah, sorry for that...
<hannes>
.oO(excuse me for removing code and rewriting chunks of code)
<mato>
hannes: btw, random meta-comment, using some kind of more advanced templating thingy to generate Makefile rather than just "append fmt" might help understanding the generation stuff
<hannes>
mato: I would be rather eager to get rid of Makefile and use some ocaml code instead..
<hannes>
but for the time being, you might be right ;)
<mato>
hannes: also, re pkg-config and OPREFIX, perhaps we really should build our own pkg-config (or pkg-conf) pre-configured with the correct paths for the opam switch
<mato>
hannes: do this in conf-pkg-config, install as opam-pkg-config
<mato>
hannes: just a suggestion...
agarwal1975 has joined #mirage
<hannes>
mato: I agree, but the burden is again not hard enough on me... and I'd think using opam2 will get rid of the pkg-config (thus I don't need to spend time implementing a pkg-config only to remove it in a couple of months). I'm also considering the BSD make a minor issue. If we cannot solve it, so be it...
<mato>
hannes: opam1 will be around for a long time until it propagates into "distros most people use"
<mato>
hannes: and i didn't mean implement our own, just steal existing pkg-config or FreeBSD pkgconf (whichever is better) and build our own copy configured with the paths we need
<hannes>
mato: if you have time to build a pkg-config, I'd be a happy alpha-user :)
<hannes>
Drup: I addressed your (not sure whether all) concerns in functoria
<hannes>
mato: and, since you're looking into 691, I'm happy to answer any questions. I believe if you've questions (concrete, or more design ones), we should discuss them in the PR (so that others can jump in and ask more etc.)
<hannes>
when it comes to makefile generation, I do a first draft in ocaml + printf, but then switch to manually hacking on the generated makefile until it works... once it works, I replay my edits and adapt the printfs..
yomimono has quit [Ping timeout: 248 seconds]
yomimono_ has joined #mirage
<mato>
hannes: I need to actually try out #691, just from the discussion it's not entirely clear to me what the "end user" workflow is. However, I probably won't have time for that tonight.
<mato>
hannes: I've added a comment in the LD PR, also not 100% sure there yet :(
<hannes>
no problem. all is good! :)
<hannes>
workflow now is `mirage configure` make depend ; make
<hannes>
where mirage configure generates an opam file with all the dependencies
<hannes>
which is good since you can configure your xen unikernel anywhere, and build it elsewhere. it of course collides with some features which look for ocaml packages during configuration time
<hannes>
(the LD stuff is carefully removed before I did 691 due to brain too small)
<hannes>
or, rephrased, 691 does contain the removal of LD as first commit ;)
<mato>
configure in the sense that you can then take that directory, run make depend elsewhere where opam is installed and things should "just work"?
<hannes>
yes
<hannes>
(whereas I don't have any clue about this opam depext stuff, depend may be tweaked for supporting that)
<hannes>
and another issue is crunch/fat/tar which - during configuration time - generate stuff -- but this is left as is for now, and needs smarter people to solve
<mato>
i think that's just a case of s/opam install/opam depext -i/ ?
<mato>
(the depext thing)
<hannes>
reading the manpage of depext, it is not immediately clear to me which package to pass (or whether I need to pass all dependencies)
<mato>
all of them, it's just a wrapper over install
<hannes>
ok... well, as you said, try it yourself, maybe tomorrow... it should just work that way :)
<mato>
will do
agarwal1975 has quit [Quit: agarwal1975]
agarwal1975 has joined #mirage
<hannes>
\o/
fgimenez has quit [Ping timeout: 246 seconds]
fgimenez has joined #mirage
fgimenez has joined #mirage
fgimenez has quit [Changing host]
fgimenez has quit []
jermar has quit [Ping timeout: 240 seconds]
<yomimono_>
hannes: I'm pointing mirage-dev at a branch I made for ocaml-tls. It has some very hacky translations for polyvars <-> errors which I don't think you'll want to merge, but do make the thing compile against the current universe
<rixed>
Hello! Is my understand correct that mirageos uses nothing from mini-os scheduler/threads ?
brson has joined #mirage
<yomimono_>
rixed: I believe that's correct, although I'm no authority on this subject. talex5 (who is infrequently here, unfortunately), avsm, or djs55 are probably the best people to ask
<rixed>
yomimono_: thanks for the confirmation. I'm still reading the code, will find out by myself soon hopefully. :)
<yomimono_>
:D
mort___ has joined #mirage
mort___ has left #mirage [#mirage]
AltGr has left #mirage [#mirage]
mort___ has joined #mirage
yomimono_ has quit [Ping timeout: 256 seconds]
mort___ has quit [Quit: Leaving.]
mort___ has joined #mirage
mort___ has quit [Quit: Leaving.]
mort___ has joined #mirage
agarwal1975 has quit [Quit: agarwal1975]
miragebot has joined #mirage
<miragebot>
mirage/master c31363a Hannes Mehnert: fix compilation for 4.02. not sure whether reduce is allowed in this module
miragebot has left #mirage [#mirage]
<miragebot>
[mirage] hannesm pushed 1 new commit to master: https://git.io/vXxgx