XSF Editor Team - 2020-06-14


  1. pep.

    jonas’, do you prefer me to merge things now or just approve them and you'll merge later?

  2. pep.

    Have you tested announcements btw?

  3. jonas’

    which things?

  4. pep.

    PRs

  5. jonas’

    yes, I tested announcements, but only to my private email

  6. jonas’

    (see the linked MR)

  7. jonas’

    would you work on editor stuff right now?

  8. pep.

    Sure

  9. jonas’

    I see

  10. jonas’

    so the things on gitlab.com are ready to be used

  11. jonas’

    right now, emails only go to my private mail address, but that can easily be changed

  12. jonas’

    if we dare

  13. jonas’

    (oh, and if we get the new sender address subscribed to the mailing list first :D)

  14. jonas’

    pep., if you want, you can play a round of "how does this new gitlab thing feel" under my supervision now

  15. jonas’

    otherwise I’ll pick a PR which can be merged to play through the pipelines with real stuff myself

  16. Zash

    Hey, based on https://mail.jabber.org/pipermail/standards/2020-May/037495.html https://mail.jabber.org/pipermail/standards/2020-June/037552.html it looks like SASL Channel-Binding Type Capability should be XEP'd but was forgotten about.

  17. pep.

    What does the sender address look like btw

  18. pep.

    oh have already merged gitlab stuff :p

  19. pep.

    I just realized

  20. pep.

    Not like it impacts the current build anyway :)

  21. jonas’

    which current build?

  22. jonas’

    Zash, I have it in the spreadsheet, but it hasn’t been tuesday yet ;)

  23. pep.

    travis/dockerhub?

  24. jonas’

    though today will be special and I’ll do editor stuff despite it not being tuesday

  25. jonas’

    pep., did you do editor things today?

  26. pep.

    no

  27. jonas’

    ok, so it doesn’t matter indeed

  28. jonas’

    I’ll switch places and then we can coordinate, there are a few things in the pipeline

  29. jonas’

    I’ll switch places and then we can coordinate, there are a few things in the queue which need to be processed, and it’d be good if we could do that together now-ish to iron out the issues

  30. Zash

    Oh, Georgs mail wasn't that long ago. Don't stress on my account. :)

  31. jonas’

    Zash, I need to test things anyways :)

  32. jonas’

    pep., so I’ll play the game with the protoxep and then I’ll show you the (temporary) ropes for a PR

  33. pep.

    k

  34. jonas’

    https://gitlab.com/xsf/xeps/-/merge_requests/4

  35. jonas’

    I think this is going to be my favourite feature: https://xsf.gitlab.io/-/xeps/-/jobs/594352928/artifacts/rendered-changes/xep-0440.html

  36. pep.

    nice indeed

  37. jonas’

    new sender address is xep-editor-pipeline@zombofant.net temporarily. If we continue to use this flow, I’ll see that we set up a proper email sender account on atlas so that we can send from editor@xmpp.org

  38. pep.

    yeah

  39. pep.

    This is also supposed to generate an archive directly?

  40. pep.

    Ah no, mr..

  41. jonas’

    not a full archive

  42. jonas’

    just the one with the changed stuff, which is also exposed in the web UI

  43. jonas’

    okay, provisional email stuff is sorted out

  44. jonas’

    I’m hitting the green button now

  45. jonas’

    man I’m excited

  46. pep.

    I read pack@main as pac man :x

  47. jonas’

    :D

  48. jonas’

    I’m so looking forward to completely removing the docker stuff in favour of build artifacts one day

  49. jonas’

    in contrast to github, you can get a publicly readable link to the latest artifact of a job on a branch trivially :)

  50. jonas’

    one thing we still need to do manually is tagging

  51. jonas’

    but I’ve got a tool for that in the pipeline, too

  52. jonas’

    it does work mostly already, but it still has some issues with some corner cases

  53. jonas’

    ah well, one mail was generated incorrectly

  54. jonas’

    I guess some problems were to be expected if you don’t clear the runner caches after messing with stuff *a lot*

  55. jonas’

    but everything else seems to have worked flawlessly

  56. jonas’

    let me put the docker image up

  57. jonas’

    ah, the docker builder was stuck. meh.

  58. pep.

    So we'll need to give commit rights to xep-attic to the CI also?

  59. jonas’

    it already has that

  60. jonas’

    xep-attic was updated correctly

  61. jonas’

    via a Deploy Key (which grants +w to exactly one repository); private key is in a protected CI variable

  62. pep.

    k

  63. jonas’

    yet another thing you cannot easily do with github

  64. pep.

    yeah

  65. jonas’

    https://gitlab.com/xsf/xep-attic/-/commit/4741034c2874583a7678caecc51418d29c87229e?view=parallel it has a bit of noise in it because of the caches I should’ve cleared, but nothing fatal

  66. jonas’

    attic is updated on eos2 now, too

  67. jonas’

    this is looking good

  68. pep.

    yeah looks good :)

  69. jonas’

    pep., wanna handle https://github.com/xsf/xeps/pull/949?

  70. pep.

    That is..?

  71. jonas’

    workflow would be to create a MR for that branch on github (by pulling it locally and pushing it to a feature branch on the gitlab repo), checking it looks OK and then hitting the green button

  72. jonas’

    huh

  73. pep.

    I'm kinda lost with what to do wrt validation in that PR

  74. jonas’

    sorry, I got confused

  75. jonas’

    I still am to an extent

  76. jonas’

    need to read up on council logs, I have it as "to merge" in my list

  77. pep.

    I guess it's blocked on adding the possibility to include validation to registrar stuff

  78. Zash

    Wasn't the vote on the version prior to the validation changes?

  79. jonas’

    https://github.com/xsf/xeps/pull/960 this would be a good next candidate

  80. pep.

    Zash, it was yes

  81. jonas’

    workflow would be to create a MR for that branch on github (by pulling it locally and pushing it to a feature branch on the gitlab repo), adding a commit which adds a revision block, checking it looks OK and then hitting the green button

  82. pep.

    doing now

  83. jonas’

    :+1:

  84. pep.

    Maybe I'll get my own fork on gitlab for the next try :p

  85. pep.

    Why is there no pipeline?

  86. pep.

    Ah because eh

  87. pep.

    not rebased

  88. jonas’

    right

  89. jonas’

    when you rebase, you’ll also have to rebase the branch on the github side to make github see that it’s going to be merged

  90. pep.

    https://gitlab.com/xsf/xeps/-/merge_requests/5

  91. jonas’

    (needs to happen before the push to master on the github side happens; only cosmetics though)

  92. pep.

    Can I push stuff to flow's branch?

  93. jonas’

    yes

  94. pep.

    I need to add a remote etc. right?

  95. jonas’

    yes

  96. jonas’

    flow is particularly nasty, because his xeps fork isn’t called xeps

  97. jonas’

    you’ll need git@github.com:flowdalic/xeps-xsf

  98. pep.

    yeah that's fine :p

  99. pep.

    I actually had him as a remote already

  100. jonas’

    pep., needs revision block

  101. pep.

    ah fail

  102. jonas’

    also I do wonder why it also rendered '440

  103. pep.

    Is that why there's also 0440 in the rendered changes?

  104. pep.

    right

  105. jonas’

    ah, I understand why, hah

  106. jonas’

    nothing to worry about

  107. jonas’

    it bases the files to include there on diff-tree against main, but in this case, your branch doesn’t contain xep-0440 yet, so it notices that it "changed"

  108. jonas’

    (and it gets the built version from the build cache)

  109. pep.

    k

  110. jonas’

    something to fix to reduce confusion, but nothing to worry about.

  111. jonas’

    can you wait a sec?

  112. jonas’

    I’d like to patch the tooling life-ish

  113. jonas’

    I’d like to patch the tooling live-ish

  114. pep.

    k

  115. pep.

    hmm I just pushed the revision block. I'll rebase when you're done

  116. jonas’

    I pushed fixes

  117. jonas’

    (to your branch :-x)

  118. jonas’

    lets see what that does

  119. pep.

    k

  120. jonas’

    huh

  121. jonas’

    ahhh... meh

  122. jonas’

    corner cases everywhere :D

  123. jonas’

    fixed

  124. jonas’

    pep., ^

  125. pep.

    "50" being a magic number that's just likely to work most of the time? :p

  126. jonas’

    it’s also the depth used by the gitlab runners by default

  127. pep.

    k

  128. pep.

    ok, CI passed for the MR!

  129. jonas’

    yupp

  130. jonas’

    it also extracted the correct file :)

  131. pep.

    yeah

  132. jonas’

    so you can hit the button from my side

  133. pep.

    from your side?

  134. jonas’

    uh.. germanism probably

  135. jonas’

    "I am ok with you hitting the green button now"

  136. pep.

    done

  137. jonas’ watches the pipelines

  138. pep.

    If this works out I hope we switch quickly so we don't "have to" maintain both in parallel

  139. jonas’

    sure

  140. jonas’

    if board approves, I intend to do a hard switch in the next two weeks

  141. pep.

    What about current PRs? rebase/push to gitlab? Grabbing authorship for all of them?

  142. pep.

    Also I'm sure some are gonna grump saying "I don't like gitlab blah blah"

  143. jonas’

    I intend to do a clean re-import of the repo once we agree on switching

  144. jonas’

    pep., oh, I have an easy reply for them

  145. jonas’

    "Are you going to maintain the build process on github in the same quality? Then do that and we stay there."

  146. pep.

    Or .. "github is more popular, we're jeopardizing the XSF's popularity!!"

  147. jonas’

    "you can send MRs via email without account"

  148. pep.

    "email is an ancient tech!!" :P

  149. jonas’

    (maybe, I haven’t checket the "without account" part)

  150. pep.

    I don't think you can

  151. jonas’

    this is perfect, everything went exactly as expected

  152. pep.

    It's certainly possible to use your github account to connect to gitlab though

  153. jonas’

    ah yes, you need an account

  154. jonas’

    and yes, you can sign in with github

  155. jonas’

    (images on eos2 pulled)

  156. jonas’

    I’m going to do the tagging for you and synced to github

  157. pep.

    hmm, we'll have to push your CI change to flow's branch on github

  158. jonas’

    do we?

  159. jonas’

    nope, all good

  160. pep.

    k

  161. jonas’

    the only requirement is that the commit ID of the head of the branch shows up on master

  162. jonas’

    doesn’t matter if it’s via a merge commit or followed by other commits

  163. jonas’

    all set

  164. jonas’

    I like this flow

  165. jonas’

    I’m going to compose an email to editor@ and standards@ now

  166. pep.

    k

  167. jonas’

    mail sent

  168. jonas’

    enough pondering

  169. jonas’

    it is also amazing that the gitlab ci runners are nearly twice as fast as docker hub, even for the non-incremental builds

  170. jonas’

    (and if we still felt them to be too slow because of a shared resource, we could host our own e.g. on eos)

  171. pep.

    Is it possible in github to prevent a repo from accepting PRs?

  172. pep.

    It's only possible to disable issues right?

  173. pep.

    We can add templates for github etc. I guess saying "New location: .."

  174. jonas’

    pep., you can archive a repository

  175. jonas’

    which is what I intend to do

  176. pep.

    I wonder if we won't be asked to compromise and mirror instead :/

  177. pep.

    not to inconviene lazy people using github, or sth :p

  178. jonas’

    would be an option

  179. jonas’

    sam also suggests to use GitLab CI on a GitHub repo

  180. jonas’

    should be possible, never tried it

  181. jonas’

    wonder how terrible the integration is going t obe

  182. jonas’

    should test that with a non-xsf repo some time next week

  183. pep.

    It is possible at all? does github integrate that?

  184. pep.

    We can surely mirror a github repo, but then we lose the CI for MRs

  185. pep.

    We can also mirror a github repo, but then we lose the CI for MRs

  186. jonas’

    pep., gitlab offers "CI/CD for External Repo" when you create a new project

  187. jonas’

    need to test that

  188. pep.

    Ah, ok

  189. pep.

    something something webhook I guess

  190. pep.

    well just like other CI solutions..

  191. jonas’

    yes

  192. jonas’

    but we might still gain the docker registry + artifacts + cahing

  193. jonas’

    but we might still gain the docker registry + artifacts + caching

  194. jonas’

    but we might still gain the docker registry + artifacts + caching + hidden CI variables + …