00:41 <haesbaert> MUAHAHAHAHAHA https://github.com/haesbaert/awa-ssh/commit/a260cdd4eac61ca9e5a0893e8b195a7178b09f03

00:41 <haesbaert> MUAHAHAHA

06:56 <reynir> Nice!

07:46 <hannes> haesbaert: nice~

08:09 <haesbaert> :D

11:31 <dstolfa> Hello, I was curious what paravirtualization features you make use of for KVM, and what would be needed in order to support bhyve natively, should one want to not use VirtIO

11:39 <hannes> dstolfa: mato (currently absent here) worked over the last week more on cutting Solo5 into pieces to allow more backends (KVM, Hypervisor.Framework, VMMAPI (the userspace API for the VMM kernel bits in FreeBSD)).

11:39 <hannes> dstolfa: I expect that there'll be patches this month for Solo5 (there's already a PR to support Hypervisor.Framework natively)

11:41 <dstolfa> hannes: Ah okay, Would this be of any help in terms of performance/usability in your design? reviews.freebsd.org/D8100

11:44 <hannes> dstolfa: thanks for the pointer, could you open a GitHub issue at https://github.com/Solo5/solo5 about "native FreeBSD vmm port" and put the link there (otherwise it will likely get lost, and I'm not the one doing low level system design)?

11:45 <dstolfa> hannes: Sure thing, thanks for the feedback

11:48 <hannes> dstolfa: thank you for the pointer. are you by any chance working on bhyve? :)

11:50 <dstolfa> hannes: On-and-off, I'm working with rwatson@ in trying to trace VMs with DTrace, and I'm doing that through bhyve. This is something I needed in the design, but I assumed it'd be useful for mirage too :)

11:51 <hannes> dstolfa: cambridge-based?

11:52 <hannes> (I just returned to Cambridge computer labs from several months in africa)

11:52 <dstolfa> hannes: Sort of, I'm studying in Croatia right now, finished my applications to Cambridge 2 days ago(MPhil already went through the department I think, PhD awaiting)

11:52 <dstolfa> hannes: Ah, I was there 2 weeks ago

11:52 <hannes> dstolfa: nice! :)

13:03 <argent_smith> hi ppl. i'm preparing a small MirageOS introduction for my colleagues at work. Question: is there something like a web page listing Mirage success stories/use cases?

13:05 <hannes> argent_smith: http://unikernel.org/blog/2016/unikernel-nfv-platform and the various papers (on jitsu, etc.) may be a good start!?

13:06 <argent_smith> thanks a lot!

13:06 <hannes> argent_smith: parts of the (ISC-licensed) libraries developed under the MirageOS umbrella are nowadays used at docker for mac and docker for windows... but I'm not sure whether they wrote anything detailed about that

13:07 <argent_smith> hannes: in my local speaks i've already mentioned these so my fellows asked me to introduce the whole thing :)

13:08 <hannes> argent_smith: http://www.databoxproject.uk/ also uses MirageOS ; http://nymote.org is an idea (vaporware atm) to use MirageOS

13:09 <argent_smith> though being an OCaml's long-timer I'm quite a noob to this area ;)

13:10 <hannes> argent_smith: and http://ownme.ipredator.se is a MirageOS unikernel which protects >10 BTC

13:45 <kensan> If I may intrude... has nymote been sunset or is it still under development?

13:47 <hannes> kensan: it's vaporware, as described. there has been ongoing work on MirageOS (and porting it to various architectures, such as kvm), on irmin as well (it now has a 1.0 release iirc).

13:48 <hannes> signpost is still vaporware (some prototype, no-longer-working scripts)

13:48 <hannes> dinosaure developed mrmime, a mime encoding and decoding libraries purely in ocaml, which works for 99% of a dataset of 2 million mail (afair, don't quote me on the numbers)

13:51 <kensan> hannes: In the introduction of your BornHack '16 talk about MirageOS you mention "hive". Is that still a blue sky idea or are you actively working on it?

13:51 <hannes> I'm actively working on blue skies.

13:52 <kensan> touché :)

13:52 <hannes> there's no release or source up in the open (yet!?) -- but in the end, MirageOS is a set of libraries, and so will projects on top of MirageOS

13:52 <hannes> atm I should be busy with some more TCP/IP testing, and congestion control.

13:54 <hannes> since nymote.org launched, we at least managed to release a tls library to-be-used with MirageOS. as mentioned, I'm currently working on some TCP/IP things, and will likely afterwards work on higher layers.. but it doesn't make much sense (for me) to build upon this brittle TCP/IP stack that is out there atm

13:55 <hannes> (others are happily using this TCP/IP stack in production, maybe it's me who is a perfectionist here)

13:55 <kensan> hannes: Ah, thats an insightful piece of info, thanks.

13:56 <hannes> kensan: as mentioned, this is my very personal opinion. docker for mac/win ships mirage-tcpip and it seems like they're happy with it.

13:56 <kensan> hannes: I am thinking about building something with MirageOS once the Solo5 Muen port is stable and ready for submission.

13:57 <kensan> hannes: Trying to get an idea of the current state of the MirageOS ecosystem that is why I am asking.

13:57 <hannes> kensan: you're lucky. the Mirage3 release is the best/most mature so far! :)

13:58 <hannes> kensan: I run various unikernel (hannes.nqsb.io is a Canopy unikernel, nqsb.io is another unikernel, I also have the authoritative DNS server for nqsb.io as MirageOS unikernel, https://tls.nqsb.io is another unikernel)

13:59 <kensan> hannes: I have been lurking but the 3.0 release and Solo5 work have made me take another, closer look and since I got a basic Muen port done in a weekend it would be a shame to stop...

14:00 <hannes> :)

14:07 <kensan> So far I have not really touched Ocaml, except for the mirage tool adjustements. Writing a unikernel to run on top of Solo5/Muen should be instructive.

14:14 <gjaldon__> hannes: where did you deploy your unikernels?

14:16 <hannes> gjaldon__: private server in a data center

14:17 <gjaldon__> ok :)

14:20 <hannes> but in solo5 there's a google cloud engine script afaics

14:22 <gjaldon__> hannes: about Canopy, I still couldn't run it locally on my arch linux I already have compiled it successfully. does it expect a cookie in git jar?

14:24 <hannes> gjaldon__: there shouldn't be a cookie needed, but I remember a similar `read` error in ocaml-git or irmin issue tracker...

14:24 <gjaldon__> would you happen to have an idea what's causing the error here: https://gist.github.com/gjaldon/83985d294ef724568a35d092726daee5

14:24 <hannes> for me, switching from smart-http (https://github.com) to git (git://github.com) solved the issue

14:25 <gjaldon__> I think I already tried that. will try again

14:26 <hannes> gjaldon__: https://github.com/mirage/ocaml-git/issues/138

14:29 <hannes> (seems to be the same issue afaics..)

14:32 <gjaldon__> I tried running `./main.native -r git://github.com/Engil/__blog.git` but get a slightly different error this time

14:33 <gjaldon__> https://gist.github.com/gjaldon/52808e7b21c8776c9f599be0900bc6da

14:35 <hannes> gjaldon__: i don't remember to have seen this before, maybe report it to ocaml-git and see whether someone has an idea there (sorry that I don't have a solution)

14:35 <gjaldon__> no prob at all hannes! thanks for your time though

16:08 <miragebot> mirage/master dc90a37 Mindy Preston: Merge pull request #799 from yomimono/fix-changelog-date...

16:08 <miragebot> mirage/master 96b6f79 Mindy Preston: put the correct release date for version 3.0.0

16:08 <miragebot> [mirage] yomimono pushed 2 new commits to master: https://git.io/vywFN