#linuxcnc-devel Logs

Jan 27 2022

01:46 AM silopolis_vm[m]: pere: fantastic! Didn't expect an immediate merge 🎉🎉🎉
01:53 AM silopolis_vm[m]: About the numbers, that looks weird... May be we should diff our POs to see why? 🤔
01:55 AM silopolis_vm[m]: This makes me think it would be nice to have some kind of board to help collaboration, what do you think? Ideally something integrated with gh.
01:55 AM silopolis_vm[m]: A self hostable one of course 😘
02:13 AM hansu: Have you seen this on weblate?
02:13 AM hansu: The translation is temporarily closed for contributions due to maintenance, please come back later.
02:13 AM hansu: The translation was automatically locked due to following alerts: Could not push the repository.
02:22 AM pere: silopolis_vm[m]: the one removed file I mentioned is not an english placeholder, so perhaps have a look at it first?
02:24 AM hansu: And also strange that Weblate only updated the files from linuxcnc.pot but not the ones from gmoccapy.pot even though both were merged into weblate/master
02:24 AM pere: silopolis_vm[m]: I've generated docs/po/Documentation-new_fr.po from the translation-po4a branch, copied it as docs/po/fr.po into my po4a-build branch, which is the one I will be pushing towards master. I plan to rename translation-po4a to translation-po4a-migration and po4a-build to translation-po4a, to get the correct branch into https://github.com/LinuxCNC/linuxcnc/pull/1356 for merging. The current translate-po4a branch will eventually be thrown
02:24 AM pere: away.
02:25 AM pere: hansu: no idea what is going on there, the github push was set up by someone else.
02:25 AM pere: Perhaps some API key expired?
02:29 AM pere: I got weblate access to open access again, but no idea if that will help, as I assume it will lock down quickly if the problem still exist. should I try to open?
02:31 AM hansu: Yeah, maybe it's worth a try
02:32 AM pere: unlocked and tried to push again, still "GitHub API access for api.github.com is not configured".
02:32 AM hansu: :(
02:35 AM hansu: But the web interface is open again
02:36 AM pere: I assume someone with commit rights to github can still do git pull from weblate and git push to github, which is my normal method of moving stuff from weblate to master. No idea how this 'create PR requests automaticalliy' stuff was set up.
02:39 AM pere: silopolis_vm[m]: <URL: https://github.com/LinuxCNC/linuxcnc/pull/1356 > is now the branch I propose to merge to master.
02:41 AM hansu: But that shouldn't be neccesary as Weblate opens pull requests automatically. I mean a manually 'copy' from Weblate to LinuxCNC wouldn't solve this as the files are not updated on Weblate
02:42 AM pere: the files on weblate are also updated manually with a button click.
02:43 AM pere: do not remember who set up the automatic PR stuff, but suspect silopolis_vm[m].
02:43 AM linuxcnc-build: Hey! build 0000.checkin #8512 is complete: Success [3build successful]
02:43 AM linuxcnc-build: Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/8512
02:51 AM hansu: >the files on weblate are also updated manually with a button click.
02:51 AM hansu: hmm but not in this case
02:54 AM pere: the weblate git repo is https://hosted.weblate.org/git/linuxcnc/linuxcnc/, if you want to have a look.
03:02 AM hansu: Yes i know, but at 07c200aad2420aed26db4bdb0e1252f9fc3ec1ac only the files in src/po were updated
03:03 AM hansu: I miss those in src/po/gmoccapy
03:03 AM pere: perhaps I was on the wrong page when pressing 'push'?
03:04 AM pere: which ones do you miss in src/po/gmoccapy?
03:05 AM hansu: Maybe.
03:05 AM hansu: de.po, es.po. fr.po, hu.po .... and so one - those I miss
03:07 AM pere: perhaps no changes were done do those since the last commit?
03:08 AM hansu: They are changes in the gmoccapy.pot if you change the commit before (the merge commit)
03:25 AM silopolis_vm[m]: pere: ok, will review deleted files...
03:26 AM silopolis_vm[m]: Got ya for branches renaming 👌
03:27 AM silopolis_vm[m]: And I don't think I was the one that enabled gh integration with weblate. See admins there, the one along us should be the man in charge
03:29 AM pere: either you, jepler[m]1, seb_kuzminsky or me, as we are the admins on weblate. I did not.
03:30 AM pere: or perhaps we got it automatically, and something broke on the weblate side?
03:36 AM pere: anyone got any idea what the error is with the html test in <URL: https://github.com/LinuxCNC/linuxcnc/pull/1356 >?
03:49 AM pere: also got no idea why the package test fail. :(
04:53 AM pere: smoe: I need help fixing the test builds on github.
05:29 AM smoe: Such stale links happen when a document was removed that offered these anchors before. Previously there were original English variants as a _es/_fr copy for that reason, I presume.
05:30 AM pere: ah, perhaps it is related to the translated templates.
05:30 AM smoe: You have many more stale ones than I ever had, though.
05:33 AM smoe: I am afraid I cannot help you over the next days. I have this https://github.com/LinuxCNC/linuxcnc/pull/1516 out for which I merged all my local changes to the English texts with the current master and then think to have addressed many other adoc formatting issues - and also removed anything that annoyed me with stale references.
05:34 AM smoe: On the more serious sides - you can remove all the _es and _fr files, cannot you?
05:34 AM smoe: Those should then no longer make any problems.
05:35 AM smoe: Yes, they will likely be unreadable for the first two (or let it be four) weeks, but then we should get somewhere.
05:37 AM smoe: Can you make -C src docs without any of these warnings?
05:50 AM pere: smoe: I did remove all translated adoc files, at least.
05:50 AM pere: smoe: I can make 'debian/configure; debuild' successfully, so not sure how to trigger the errors.
06:21 AM smoe: pere : in .github you see the exact commands that are executed for the CI checks
07:33 AM silopolis_vm[m]: pere: Master_Integrator_fr.adoc was moved to docs/archive/fr/ because it was to much different from en version: english has 3 included files whereas _fr had 10s
07:33 AM pere: believe I found the bug. DOC_SRC -> DOC_SRCS in Submakefile....
07:33 AM pere: silopolis_vm[m]: ok. glad to hear it was expected.
07:34 AM silopolis_vm[m]: I think back then I checked that all files where included elsewhere in en docs
07:35 AM silopolis_vm[m]: I may have failed in some places, but have been very careful not to loose any content in the process
07:36 AM pere: is the PO file you ended up with very different from the one I generated?
07:38 AM smoe: I admittedly do not want to check. The English texts have (syntax-wise) now improved. I have no idea about how stable po4a and weblate are.
07:38 AM silopolis_vm[m]: pere: I get yours from the po4a-build branch right ?
07:38 AM smoe: I should say "robust"
07:39 AM pere: silopolis_vm[m]: kind of, but it was renamed this morning.
07:41 AM smoe: I would expect so. Do not think I ever committed the .po files into any of my branches.
07:44 AM smoe: We should possibly have synced this more with seb_kuzminsky. The fear of ours was that the .po effort would not be mergeable if we correct the syntax too much. But if the translations got this right, then we cannot just go and make the translation worse to get a sync. The quality improvements that come with the .adoc review and all the changes to
07:44 AM smoe: the docs that are admittedly difficult to review in github, this is something that we should have planned more.
07:46 AM pere: well, we are moving to a state where the english text is the authorative source, so if the translations had it right, they got it wrong by not pushing the improvements back to english.
08:21 AM smoe: That is what silopolis_vm[m] just did. I was also annoyed. Very often. But then again, the translators are commonly native speakers and prefer writing in their own language. I said this elsewhere already - the docs need special attention and we yet do not have the infrastructure for them.
08:21 AM smoe: Should there be a #linuxcnc-docs channel?
08:23 AM smoe: Also, just -docs is a bit retro. We could also think of tutorial videos and transcriptions/subtitles for those in various languages.
08:48 AM pere: I do not know the content enough to have an opinion on those things, but I am sure those are good features to have in place. My focus with the po4a change is only to make life easier for translators while ensuring a consistent set of documents across languages. I do not try to improve the quality of the documentation at the same time, it must be done separately.
10:35 AM seb_kuzminsky: i merged the hal_init fix to 2.8, merging it to master now
10:53 AM seb_kuzminsky: ok it's merged in master too
10:53 AM seb_kuzminsky: whew!
11:08 AM smoe: congrats!
11:09 AM silopolis_vm[m] is now known as silopolis[m]
12:03 PM silopolis[m]: Damn it can't reply to messages anymore in element 🤬
12:52 PM silopolis[m]: @all smoe: I'm all in with you about sequence and planning and was disapoin^Wsurprised by the suden rush into po4a/weblate. I'll also forever miss the Crowdin translation UI... But let's move on!... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/cd818aafde6b33d1743a1a315ea80be19938a3ff)
01:46 PM smoe: silopolis[m]: I think we are pretty much in sync. Concerning weblate or alternatives I am completely blank. So far I am a bit disappointed about the weblate community, but guess what, it is not easy. And quite a few demand changes to the English, one tends to think. But that is also why we are all doing this. I was surprised about a few instances
01:46 PM smoe: for which the translation does not fit at all. This is dangerous if it goes unnoticed.
01:47 PM smoe: The challenge for weblate and the long texts are - the long texts. At least I hope that we do not have to translate every bullet individually.
01:53 PM silopolis[m]: Long texts are a challenge for everyone, final reader included!
01:53 PM silopolis[m]: docs/src/ should hold nearly twice the number of files!
01:54 PM smoe: What I anticipate is that automated translations will be used a lot and then checked for correctness. pere pointed to the problem that this might violate copyrights, I just do not know about the conditions that the "suggestions" in weblate may be used.
01:55 PM smoe: DeepL in particular does a good job, so I find.
01:55 PM pere: smoe: At least I find the copyright status of machine translated text quite unclear. is it even covered by copyright? If so, who is the copyright holder...
01:55 PM silopolis[m]: <smoe> "The challenge for weblate and..." <- That 's why I invested so much in saving french translation and by respect for contributors before us
01:57 PM silopolis[m]: smoe: J, that's something that never came to mind 😱😬
01:58 PM smoe: pere expect that weblate has answered this prior to offering those buttons. But we should ask. As long as we can modify and redistribute what is suggested, I do not really care, I must admit. Just list DeepL and Google Translate as contributors.
01:59 PM pere: smoe: why do you expect that? I suspect it it is a legal phd question to figure out the answer to that one.
01:59 PM smoe: I am a bot, too. At least this is what my neighbours think when they see me at the computer the whole night.
02:01 PM smoe: Ok. Assume we would have a file README_bark.adoc and my dog would do the translation. Who owns it?
02:01 PM pere: no-one. docs can not hold copyright.
02:02 PM smoe: Why not? And if the dog cannot hold it, why should a machine?
02:02 PM pere: already texted in court, a photographer claimed copyright for selfie taken by monkey and lost. someone else claimed copyright for the monkey and lost too.
02:02 PM pere: s/texted/tested/
02:03 PM pere: only natural persons can create 'works' with the current copyright laws.
02:03 PM smoe: Then why exactly are you worried about the automated translations?
02:04 PM pere: the automated translations are derived from translations created by natural persons, so they are derived works, which complicates matter a lot.
02:04 PM silopolis[m]: Anyway, that's not translation IA's business model to copyright strings that, anyway, are only suggestions made to translators so that she can improve... IA!
02:05 PM smoe: If we are allowed to use their services and are not sneaking/hacking their system, then it should be fine. It is not like we are hiding anything. And likely they use our translations for their training, too.
02:05 PM smoe: Again - this is on weblate.
02:06 PM silopolis[m]: They don't give a shit about (c), they want us to fead the beast freely!
02:26 PM silopolis[m]: smoe: anything to review in your PRs to master? Lost track, are they all merged?
02:28 PM smoe: The English-documentation-ones is in the open, still: https://github.com/LinuxCNC/linuxcnc/pull/1516
02:28 PM silopolis[m]: Do we agree on my informal proposal above? How do we proceed next?
02:29 PM smoe: pere This one https://libera.ems.host/_matrix/media/r0/download/libera.chat/cd818aafde6b33d1743a1a315ea80be19938a3ff I think I agree with
02:31 PM silopolis[m]: Should we try to merge my en edits to your branch to have a single PR to master?
02:31 PM smoe: Prior to the upload to weblate as I understand it the English bits need to be updated. In what sequence - I do not care.
02:33 PM smoe: silopolis[m] I think we should ask seb_kuzminsky about how he prefers this. seb_kuzminsky, if you are around, is there are plan from your side?
02:34 PM silopolis[m]: My concern is really ease of merge, git history quality and PR acceptability
02:34 PM silopolis[m]: smoe: U read my mind 👍
02:34 PM jepler[m]1: I'm trying to figure out what's up with weblate but I don't know yet. It's supposed to be creating pull requests automatically, and I don't know why it's not.
02:41 PM jepler[m]1: I manually took the accumulated changes from weblate and pushed them to github.
02:42 PM jepler[m]1: I still don't know what's up, I verified every element of the config on hosted weblate against circuitpython, where pull requests are created as expected..
02:49 PM linuxcnc-build: build #8516 of 0000.checkin is complete: Failure [4failed fetch branch to local git repo] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/8516 blamelist: Sebastian Kuzminsky <seb@highlab.com>
03:00 PM silopolis[m]: Just to let yall know, I'm currently importing my gh fork of linuxcnc into https://taiga.io to have a look at it...
03:01 PM jepler[m]1: https://github.com/LinuxCNC/linuxcnc/pull/1539 OK I went and changed some french translations based on deepl suggestions and it DID create a pull request on github this time... maybe it'll behave in the future? of course, the pull requests need to be approved & merged by someone.
03:02 PM jepler[m]1: we can give additional folks review & merge access even (or especially) if you would use it just for allowing translation work to enter to start with,.
03:03 PM silopolis[m]: Here https://tree.taiga.io/project/silopolis-silopolislinuxcnc/timeline
03:05 PM silopolis[m]: jepler[m]1: That would surely help smooth the process and relieve mighty devs from this chore 😘
03:09 PM silopolis[m]: <silopolis[m]> "Here https://tree.taiga.io/..." <- I'll make tests this we to see how it really works and if it can help us. @smoe @pere or any other tell me if you need creds to test too, and please shoot out any testing scenario coming to mind 🙏
03:16 PM silopolis[m]: smoe: plan for the we also includes reviewing and commenting your coding guidelines PR. But I still have a bunch of adoc docs reading to do...
03:18 PM silopolis[m]: pere: anyrhing to put on my table before all this?
03:20 PM pere: silopolis[m]: I'm currently busy elsewhere, and if I find time for linuxcnc I will try to fix the po4a HTML build which has issues. do not think I have anything for your table.
06:39 PM smoe: silopolis[m] : Please invite me ot this taiga thingy.
06:42 PM smoe: The link above somehow does not work.
10:39 PM -!- #linuxcnc-devel mode set to +v by ChanServ