hannes changed the topic of #mirage to: next bug cleaning day: 2018-01-05 14:00 UTC - late (every first friday of a month); MirageOS 3 is released, happy hacking!
argent_smith has joined #mirage
<tg>
well, they could develop some antispam features to prevent this crap.. there's this mass_hl_blocker script that helps a little bit but not much really..
<tg>
could try banning nicks ending with numbers though.. but of course irc is too dumb to do that properly.. no regex only * ? wildcards
<tg>
maybe try something like +b *0*!*@* *1*!*@* etc and add +e exceptions for the few nicks with numbers
<tg>
and/or add some spam detector bot that kickbans these things right away
<Drup>
solving that as a the channel level is not a great idea
<Drup>
it's a freenode-wide issue, might as well use freenode tools
<h01ger>
can you please take this discussion elsewhere?! for me, its more annoying that those spammers, as those are trivial to ignore, while if you say stuff, i let my mind proceed it
<reynir>
Yes
<hannes>
morning!
<h01ger>
moin hannes
<reynir>
morning o/
niks3089 has joined #mirage
yomimono has joined #mirage
<lobo>
oh, i'm one hour early. enough time to get a new 0xc0ffee
<yomimono>
perhaps "morning" is the time at which coffee is made, which for hannes, is all times :)
* avsm
drinks his coffee thoughtfully
<yomimono>
at any rate, let me refresh this here agenda...
TImada has joined #mirage
<yomimono>
first up, I had a reminder about bug-cleaning day, which is the first Friday of every month. This month, that's this Friday!
<avsm>
wow!
<yomimono>
we'll coordinate here starting from around 14.00UTC until sometime later, but don't feel too constrained by that if you want to participate :)
* hannes
is in berlin, but still lacks sleep from 34c3...
<avsm>
so, i had a suggestion about this -- I've cleaned up the ocaml discussion forum at https://discuss.ocaml.org, and there is now an Ecosytem category and a tag for `mirageos`. So one good use of the forum would be to shift longer discussion items from GitHub onto that forum, where it's easier to reference from multiple github places
<avsm>
there is even a discourseforumbot that i'm investigating that posts to a github issue whenever its referenced from a Discourse thread
<yomimono>
So I think it's reasonable to put new discussions there, but not to try to redirect conversations that have been ongoing in issues for a long time there, as it can make conversation hard to follow
<Drup>
avsm: I like the new organization a lot
<avsm>
yomimono: yeah -- this might make more sense for new issues or for library discussions
<avsm>
Drup: great!
<yomimono>
it's quite likely that we'll come across at least one cross-cutting issue or idea on Friday to seed some discussion with ;)
<hannes>
since we managed to write up our experiences from december, there's another one in march (7th - 18th). all creatures welcome!?
<hannes>
so please sign up, it will be slightly longer than the last ones (spanning from wednesday till the sunday next week)
<avsm>
at least i know the way to the place from the market now
<hannes>
if anyone would like to participate only some part of the time, please tell. we'll find a solution. i hope there'll be enough beds for everybody :)
<hannes>
that's all i have to say... if there are any questions, don't hesitate to reach out to me via whatever channel you prefer
<avsm>
thanks for organising this again, hannes!
<yomimono>
thanks hannes :) the hack retreats are great and we appreciate all your work to make them happen!
<hannes>
(and as another spoiler: i'll participate bornhack.dk and help them with the network by running dhcp and dns services (mirage/solo5/freebsd))
<reynir>
yey
<yomimono>
I would really like it if we could do something especially nice for the people who run Priscilla this time - both for hana and siham and for the people who work in the kitchen and do the cleaning
<hannes>
it's ~200 people camping for a week on bornholm in summer
<avsm>
yomimono: maybe a working internet uplink :)
<h01ger>
yomimono: one idea was to maybe bring them tools
<h01ger>
non-IT tools…
<hannes>
an apu with dualsim and 4G and load balancing plus some wireless access points would be good to get there (i'll try to get that, have friends in berlin who know that stuff)
<yomimono>
it's a bit hard for me to organize that as NA is extremely inconvenient to bring objects to Marrakesh from - I usually get ~3 security checks from my door to Priscilla's and each object tends to be inspected with great interest :)
<yomimono>
so I am hoping a more convenient volunteer will appear and accept my heartfelt thanks
* h01ger
nods
<hannes>
yomimono: we can coordinate on the mailing list... there's already one guy who signed up via mail :)
<avsm>
pretty easy from cambridge -- can procure
<yomimono>
hannes: good point, mailing list is the place to discuss this, sorry for the derail here :)
<yomimono>
anything else on the hack retreat?
<yomimono>
...then next up, we had niks3089 with an update on core dumps in solo5/ukvm :)
<yomimono>
tell us all about it, niks3089!
<niks3089>
Yes, I've been working on a way to generate a core dump to inspect when an error occurs in Unikernel. I was able to do this with solo5 and ukvm (With the help of Ricardo).
djs55 has joined #mirage
<niks3089>
So when a trap occurs, the kernel part of solo5 will communicate the register info to the ukvm using a hypercall and using this information a core file is generated which can be used to debug what went wrong.
<avsm>
nice -- is this using the gdbserver protocol?
<niks3089>
No
<mato>
it's just an ELF core dump, right?
<niks3089>
yes
<avsm>
ack
<niks3089>
currently works with linux/x86/ukvm
<niks3089>
I've also added solo5_assert() API which the application could use to dump the core when an assert situation is encountered.
<mato>
nice. can you submit an initial PR for review and we can discuss it there?
<niks3089>
Yes, I am fixing a bug after which will create a PR
<mato>
sounds good
<yomimono>
looking forward to that! thanks, niks3089!
<yomimono>
if that's all --
<mato>
i'm about half way through reinstalling the box the solo5 CI runs on so that i can actually run full end to end tests from the CI in nested virt
<yomimono>
oh *nice*
<mato>
since the current CI is just "it builds"
<avsm>
ive got a physical box for new mirage.io servers now too
<mato>
(well, and qemu software emulation for the virtio tests but that doesn't really count)
<avsm>
just need to debug the 3.0.8 bug on dynamic block devices in xen
<avsm>
will have time on the weekend i think; this weeks a bit full already
<yomimono>
avsm: it's 100% my fault, I broke it
<mato>
i should have the (linux surf-build) CI back up by the end of the week
<yomimono>
I'd be happy to rotate it if we had another (even a simpler, don't care) test that tells us whether Xen still works
<hannes>
and get the let's encrypt stuff up and running really soon now (but looked into package signing last days)
<avsm>
yomimono: agreed, will keep for now
<mato>
we should have both
<yomimono>
mato: agreed, that would be better
<mato>
and just have two A records for mirage.io
<mato>
no need to go overboard with load balancers and such imo, the point is just to dogfood both backends
<avsm>
yeah
<hannes>
mato: +1
<avsm>
thomasga was going to look at simplifying the website build a little with ramen as well
<mato>
we could just migrate the website to canopy (or something like it) *hides*
<yomimono>
I think that's been avsm's plan for a long time
<yomimono>
although I should let him speak for himself :)
<avsm>
just been waiting on some of the web libraries to mature; i released yaml over xmas and thomasga's website is using ramen now (and mine soon too)
<yomimono>
I don't have a particular attachment to the website as a particular cog in our CI, but I do have an attachment to having any kind of automated test of whether MirageOS is currently working
<thomasga>
yea it was my plan to (and it still is)
<yomimono>
so I don't want to lose the only thing we have that actually boots stuff without replacing it with something else that actually boots stuff
<thomasga>
(refreshing + canopy-like style)
<thomasga>
but the flu was stronger that my plan :p
<avsm>
yomimono: agreed
<yomimono>
oh no :( hope you're feeling better, thomasga, that stinks
<yomimono>
eat lots of ramen ;)
<thomasga>
thanks I will :-)
<avsm>
oh last thing! If anyone has time to create an "About MirageOS" topic for the Ecosystem category on discuss.ocaml.org, it would be very useful
<avsm>
just a simple thing with pointers to the website and what it is -- it can then show up in the central FAQ on that site
<reynir>
Yes. IIRC the handler would get "echo source-vm" instead of "echo" as the command. Sorry, I can't remember exactly
<hannes>
I btw recently stumbled upon yomimono at https://media.ccc.de/v/34c3-8949-library_operating_systems - a very nice talk (the recording I listened to hat sometimes lacking audio, but i reported that, and it may be fixed at a future point -- worth watching, great talk :D
<olle>
o that's grand!
<yomimono>
it was translated into german and french and perhaps that audio is better? click the gear to try it, if you speak either of those languages :)
<reynir>
Also, qvm-run now (since when?) sends command »QUBESRPC qubes.VMShell source-vm« and sends the actual command as payload. IIRC it would send »echo; exit« or so - not very nice
<yomimono>
thanks hannes :) it was a pleasure, lots of good conversations both after that and at the mini-assembly.
<reynir>
talex5: No, that issue is unrelated, and I'm not sure if they still send nogui: - I couldn't reproduce last time I tried
<yomimono>
I agree (with hannes, I think) that we should consider an ocaml/mirageos assembly for 35c3 :)
<reynir>
yomimono: Yes, there were quite a lot of people at the Mirage meetup !
<yomimono>
(btw, if you were looking for an official end to the meeting, I declare it to be now)
<yomimono>
(partly selfishly, because my computer and I need to go for a walk! thanks folks, see you Friday for bug cleaning :D)
<reynir>
o/
<hannes>
thanks for running the meeting yomimono!
yomimono has quit [Quit: Leaving]
<lobo>
bye. it was great meeting some of you at 34c3