Hi, Here's the summary of the IRC meeting.
--- COMMUNITY MEETING Place: #openvpn-meeting on irc.freenode.net Date: Thu 18th June 2020 Time: 20:00 CEST (19:00 UTC) Planned meeting topics for this meeting were here: <https://community.openvpn.net/openvpn/wiki/Topics-2020-06-18> Your local meeting time is easy to check from services such as <http://www.timeanddate.com/worldclock> SUMMARY becm, cron2, dazo and mattock participated in this meeting. --- Talked about the status of OpenVPN 2.5: <https://community.openvpn.net/openvpn/wiki/StatusOfOpenvpn25> Cron2 will continue merging plaisthos' patches, then look at tap-and-netmask-and-IPv6 issues on windows and then the VRF patch. No progress was made on reviewing the man-page patches. Dazo has some additions in the pipeline for it already. Mattock was able to produce a tap-windows6 MSM (~installer) today, so he will move forward by creating the OpenVPN MSI installers. --- Talked about automating OpenVPN MSI builds. The current Vagrant setup has a linux VM for producing the build artifacts (with openvpn-build). Those artifacts are then shared via Samba on the Windows packaging host, which then produces the MSI packages. So the automation difficulty factor is bigger than with our current "cross-compile on Linux with openvpn-build" approach. Mattock will gauge the difficulty of automating the MSI build process after he has a good grasp of the process. --- Froze the feature set of OpenVPN 2.5. The ones on "must have" list now will be delivered, everything else will be postponed: <https://community.openvpn.net/openvpn/wiki/StatusOfOpenvpn25> --- Talked about pkcs11-helper patching and upgrade for Windows installers: <https://github.com/OpenVPN/openvpn-build/pull/172> <https://github.com/OpenVPN/openvpn-build/issues/168> One option is to upgrade from 1.22 to 1.23 and use the latest Fedora patch. We could also move directly to 1.26 - the patch does apply with some offset warnings and building pkcs11-helper still works. Dazo sent email to fedora-devel mailing list and ask why Fedora is still using / is stuck on pkcs11-helper 1.22. Meanwhile mattock will produce OpenVPN 2.5 Windows installers that bundle 1.26 with the latest Fedora patch. Also noted that we can release updated pkcs11-helper in a 2.4.x Windows installer release if the new version looks solid. -- Full chatlog attached
(21:07:36) mattock: did everyone fall asleep already? :) (21:08:04) cron2: 2.5 first :-) (21:08:10) dazo: Perhaps tie this with https://community.openvpn.net/openvpn/wiki/StatusOfOpenvpn25 (21:08:12) vpnHelper: Title: StatusOfOpenvpn25 – OpenVPN Community (at community.openvpn.net) (21:08:20) mattock: I can start (21:08:23) cron2: I've merged 3 of 5 from the plaisthos patchset, and then got distracted by workers in the house (finisehd now) (21:08:27) mattock: ok go ahead (21:09:02) cron2: so, merge the remaining two, then go and look for tap-and-netmask-and-IPv6 issues on windows, and the VRF patch (21:09:17) cron2: ordex is moving to a new flat this week, so, busy (21:09:23) dazo: yeah (21:09:41) cron2: lev__ is still missing (21:09:45) dazo: cron2: did you have a chance to look at the man page stuff? Or should I just start to send patches to the ML? (21:10:18) cron2: dazo: only the look from last week, no thorough review yet (21:11:06) dazo: I see there might be some man page updates in the queue as well ... so this needs some careful coordination to ensure those additions doesn't get lost (21:12:18) cron2: I won't merge any man-page related stuff (21:14:08) cron2: anything from wiscii yet? (21:14:15) dazo: nope (21:14:48) dazo: Only that he forgot to checkout the right git branch ;-) (21:15:17) cron2: oh, and that gitlab kicked him, right :) (21:16:56) cron2: so, mattock, how's 2.5 coming along? (21:17:02) dazo: ahh, right ... for the pull-req ... well, I'm willing to grab patches sent to the mailing list (21:17:06) mattock: quick update from me: I was able to produce a tap-windows6 MSM (~installer) today, so I will continue with the MSI installer (21:17:35) cron2: can you - if it succeeds - integrate it into buildbot so we can get msi snapshots of "master"? (21:17:44) dazo: +1 (21:17:55) mattock: good luck with that (21:18:09) mattock: might be possible, assuming the Microsoft signing service has an API (21:18:16) mattock: well (21:18:20) mattock: for openvpn, maybe (21:18:35) cron2: oh, .msi needs to be signed by microsoft? (21:18:38) mattock: it will be tricky because MSI packaging will happen on a Windows host (21:18:52) ***cron2 trusts mattock's insane windows python scripting abilities (21:19:07) cron2: (talking about openvpn.msi, not tap6.msm, yes) (21:19:12) dazo: hmmm pity (21:19:17) mattock: I mean, I would love to automate it, but it will be even more challenging the openvpn-build/windows-nsis (21:19:27) mattock: s/the/than (21:19:48) cron2: I am full of trust that you will do this excellently! (21:19:50) mattock: doable, but I would not hold my breath (i.e. sometime before 2.5.0 _maybe_) (21:19:50) dazo: okay, so occasional MSI/MSM builds is the best we can do for now (21:19:59) mattock: for now, yes (21:20:04) cron2: that would be very good to give people stuff to test (21:20:16) cron2: (without you having to build a new msi every few days) (21:20:29) mattock: the vagrant setup has a linux VM for producing the build artefacts which shares them via Samba on the Windows packaging host (21:20:53) mattock: so, that could be replicated on more static instances, but yeah, not trivial to do (21:21:03) dazo: cron2: when you've merged the last patches from plaisthos, I'll kick of another Fedora Copr build for Fedora/RHEL/CentOS ... the RPM packaged openvpn-git-master builds (21:21:06) mattock: yeah, I also don't want to keep on building MSI packages every two days :D (21:21:34) cron2: dazo: yep. And ecrist can do a new freebsd snapshot :) (21:21:39) mattock: I'll try to get an idea of the complexity after I go through the process manually (21:22:18) dazo: But regarding the list of items for 2.5 ... should we now draw the line of what goes into 2.5 and move the rest to "a future release", just so we know exactly what we will focus on? (21:22:44) mattock: I think we have to (21:23:07) cron2: I think the "must have" list is it - the "try to make it happen" is too complex (21:23:15) cron2: except maybe "update auth-user-pass docs" and "Linux VRF" (21:23:35) dazo: yeah, that's what I was thinking as well (21:23:54) cron2: I'll update (21:23:57) dazo: thx! (21:25:05) cron2: like this? (21:26:44) dazo: I would also change the "we wanted" headline to "Postponed items" or something like that. To really make it clear we're closing the dev cycle (21:27:36) cron2: Postponed items (former "nice to have" items for 2.5) (21:27:53) dazo: LGTM (21:27:55) dazo: thx! (21:29:16) mattock: good on 2.5? (21:30:08) cron2: I have much work and little to say (21:30:15) dazo: think so, yes (21:30:24) mattock: pkcs11-helper + openvpn-build? (21:30:26) cron2: dazo: can you poke lev__ to answer plaisthos' mail on the compression patch? (21:30:33) dazo: I'll do that (21:31:18) dazo: (I responded a bit earlier today he was out driving at this time) (21:31:25) dazo: *He (21:33:38) mattock: https://github.com/OpenVPN/openvpn-build/pull/172 (21:33:41) vpnHelper: Title: replace rfc7512 URI patch with latest version in Fedora by becm · Pull Request #172 · OpenVPN/openvpn-build · GitHub (at github.com) (21:33:58) mattock: that's the "pkcs11-helper" issue (21:34:09) becm: (1st half) (21:34:12) cron2: have we not discussed this before? (21:34:33) mattock: "open question would be pkcs11-helper version bump (1.23+) to support tokens with EC-keys (issues/168)" (21:34:57) mattock: https://github.com/OpenVPN/openvpn-build/issues/168 (21:34:59) vpnHelper: Title: included libpkcs11-helper-1.dll is compiled without Elliptic Curve support >=2.4.5 · Issue #168 · OpenVPN/openvpn-build · GitHub (at github.com) (21:35:08) dazo: If the pkcs11-helper upgrade is just to upgrade the library and rebuild openvpn, I see no harm (21:35:32) dazo: but if we need to adopt our pkcs11-helper implementation, then it would need to wait (21:36:26) becm: and the rfc7512 patch would no longer be identical to Fedora (21:37:03) mattock: does the fedora patch merge cleanly or cause merge conflicts? (21:37:12) mattock: in pkcs11-helper 1.23 that is (21:37:17) dazo: eww (21:37:20) becm: simple bugfix for issues/168 would be pkcs11-helper 1.23 (which nobody else uses) (21:37:45) becm: mattock: off by 1 line (21:37:58) dazo: I see that Fedora and EPEL-8 builds have moved to 1.22 (21:37:58) becm: copyright-notice changes (21:38:05) mattock: ok but nothing major (21:38:38) mattock: we were already speaking of using a static copy of the patch "initially" (read: for several years if not forever) (21:38:47) becm: other distributions use 1.25.1 (OpenSUSE, Debian stable) (21:38:47) uipko [~ui...@82-94-53-40.ip.xs4all.nl] è entrato nella stanza. (21:39:24) uipko ha abbandonato la stanza. (21:39:27) dazo: Hmmm ... https://bodhi.fedoraproject.org/updates/?packages=pkcs11-helper (21:39:31) uipko [~ui...@82-94-53-40.ip.xs4all.nl] è entrato nella stanza. (21:39:54) uipko è ora conosciuto come uip (21:40:32) dazo: Hmmm ... https://packages.debian.org/buster/libpkcs11-helper1 1.25 indeed (21:40:34) vpnHelper: Title: Debian -- Details of package libpkcs11-helper1 in buster (at packages.debian.org) (21:40:43) becm: Fedora/RH stayed on 1.22 for 2 years now (21:41:48) becm: leading to the same token problem with EC-keys for OpenSSL 1.1.x (21:42:27) becm: so, the question is, do we care and can anybody test regressions. (21:42:41) dazo: right ... I see 1.26 got released in January, including a patch from Selva (21:42:43) dazo: https://github.com/OpenSC/pkcs11-helper/releases (21:42:45) vpnHelper: Title: Releases · OpenSC/pkcs11-helper · GitHub (at github.com) (21:43:40) mattock: becm: I think we do care, but we can't really test (21:44:04) mattock: but 2.5-beta would be a better place to break things than 2.5.x (21:44:12) mattock: or some random windows installer release (21:44:17) dazo: agreed (21:44:37) mattock: so would 1.26 - in theory - solve all our problems? (21:44:58) dazo: not the rfc7512 issue (21:45:23) mattock: oh yes that is the "upstream does not consider it an issue" thing, right? (21:46:01) dazo: yes (21:46:32) mattock: anyways, I'm fine with upgrading pkcs11-helper and applying the new patch (assuming it applies cleanly to 1.26) (21:46:36) mattock: fedora patch (21:46:53) mattock: and would prefer to do that a.s.a.p. so that I can get it to the MSI installers soon (21:46:55) becm: 1 line dif == cleanly? (21:47:15) mattock: I think misunderstood you (21:47:22) mattock: did not actually try to apply it (21:48:41) becm: Fedora patch -> clean for 1.22 (21:49:23) becm: everything newer, 4 warnings due to line offset (copyright change) (21:49:27) dazo: Just started a local testbuild based on the fedora rawhide .spec, updating to 1.26 ... it the patch applied with "patch offset warnings" only (21:50:07) becm: the upstream OpenSC/pkcs11-helper/pull/4 should apply cleanly (21:50:27) mattock: mm, the patch is actually quite big (21:50:32) dazo: it is (21:50:33) becm: dwmw2 does/did regular rebases (21:51:24) dazo: dwmw2 is a good guy (21:51:58) becm: maybe he knows why Fedora is stuck on 1.22? (21:52:39) dazo: He definitely does ... we just need to reach out. I can send a mail to fedora-devel asking why Fedora is stuck on 1.22 (21:53:01) mattock: sounds good (21:53:09) mattock: so we wait until we get a response? (21:54:08) dazo: I think we can plan for 1.26 in the windows builds at least, once I've seen that my test build works fine ... The patching issues I see are just these ones: https://termbin.com/l5gi (21:54:27) mattock: same here, I also tested it (21:54:41) becm: yep, looks familliar :) (21:54:43) dazo: Otherwise it builds and packages fine for me (21:56:00) mattock: so, do we want an answers to the "why is Fedora stuck on 1.22" first or shall I just go ahead and upgrade libpkcs11-helper and the patch? (21:56:06) mattock: 1.26 + patch (21:57:02) dazo: lets do it in parallel ... we start testing 1.26 for Windows and ask in Fedora (21:57:05) mattock: ok (21:57:46) becm: is there a plan for 2.4.10? (would also be a target for updated patch) (21:58:18) mattock: not afaik (21:58:52) mattock: this could be released in a Windows installer release, but then again, something could break (21:58:53) cron2: becm: well, that's more a "new windows installer of 2.4.9" if there is no actual openvpn change, just "packed libraries" (21:59:11) cron2: but we could certainly do that if testing turns out that this is good (21:59:21) dazo: yeah (21:59:43) becm: if I remember correctly, the current state in Fedora in fedora is a "frankenpatch". Headers from current upstream, content from 1.22 (22:01:17) becm: (git commit IDs) (22:02:29) mattock: anyhow, 2 minutes overtime, getting late and I think we have a plan (22:02:30) mattock: agreed? (22:03:59) cron2: yeah (22:04:09) mattock: ok, let's end this thing (22:04:15) mattock: I will send a summary in a few minutes
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Openvpn-devel mailing list Openvpn-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openvpn-devel