Conversation
|
should mirleft/ocaml-nocrypto#44 also be in this list?? |
|
@hannesm you're right - I didn't realise that still wasn't merged. Is there a problem with it? |
|
OK, I think we have all the required libraries in now! I added a Travis check to try building mirage-skeleton with |
|
mirage-console 2.1.1 in ocaml/opam-repository#3445 |
|
mirleft/ocaml-nocrypto#44 can be crossed off the list too. |
|
@pqwy - not yet - it hasn't been released yet. |
Required by latest TLS, which is required by latest conduit support.
The latest release was newer that our 0.7.1 version.
Mirage 2.2.1 is out.
|
tcpip.2.2.3 is now released, so checked that one off. |
|
I've added a ctypes.0.4.0 working version in #55, as that's necessary for the Xen compilation. |
|
I believe mirage-entropy needs to be somewhere in here... and to get this released, mirage/mirage-platform#120 needs to be merged and released first.. |
|
mirage-platform 2.2.2 packages going out in ocaml/opam-repository#3667, so mirage-entropy is releasable now. |
|
there is no mirage-types-lwt-2.3.0? |
|
am just adding it now! |
|
ocaml/opam-repository#3694 has mirage-types-lwt.2.3.0 |
|
Seems to me that mirleft/ocaml-nocrypto#44 was released. Checking it off this list. |
|
Seems various APIs had changed. I've updated the mirage and conduit PRs to the new APIs and removed the released components from this repository. Should be working again now. |
|
getting there... ocaml/opam-repository#3770 |
|
I've checked more boxes, the missing bits are cleaning-up out TLS support for |
|
That last bit is being tracked here. |
|
once mirage mirage/mirage#394 PR is merged, and there are some releases (nocrypto, tls, mirage-entropy-xen), we'll need gmp-xen and zarith-xen; which might both be sensible to drag in via a |
|
Indeed, we need the |
|
would also be nice to release a |
|
Almost done now, I've merged most of the things in
|
|
Ha yes good point! Indeed, I plan to run a bit of more tests and then we can do a big pull-request to opam-repository with a coordinated announcement and blog post. |
|
this can be closed as well, or? |
|
The extra tests added to |
|
@talex5 can you create a new PR with your travis improvement? thanks! |
|
Thanks all for the efforts, we are almost seeing the light :-) |
|
Just one thing I'm missing for the record: mirleft/ocaml-tls#250 (comment) -- will release this to get it working with cstruct 1.6.0 (@samoht tells me its cstruct 1.5.0 only atm) |
|
this PR in tls is intentionally not merged, to not introduce a dependency on |
|
also, keep in mind that tls-0.5.0 already works flawlessly with cstruct-1.6.0! |
|
Congrats, everyone! This turned out to be soooo much hairier than initially expected... 😄 |
When the Vfs code got split out from the 9p support, the unique inode ID wasn't included. Instead, the 9p support started minting a new ID each time it was asked for one. This meant that clients saw a new inode number each time they asked for it. This confused cp at least, which checks that the inode number is the same before and after opening the file. Fixes mirage#52. Signed-off-by: Thomas Leonard <thomas.leonard@docker.com>
This builds and tests everything with all the TLS patches in. I assume we now wait for these patches to be accepted upstream and update to the release archives as they happen?
Things needing to be merged/released:
And, afterwards: