Tuesday, 2019-09-17

*** zbenjamin_ is now known as zbenjamin01:33
dcalisteHello chriadam, how are you ? Thanks for the last reviews and merges.06:58
chriadamhi dcaliste, I'm well thanks06:59
chriadamhow was your week?06:59
dcalisteFine, I destroyed my shared calendar yesterday, have to restart from backup ! I synced with a small window which removed my events on device and then I resync with a bigger window which result in deleting the events on servers because deleted events on previous synced were seen as local deletion…07:02
chriadamhmm.  I see the issue.  we need to mark those events as "deleted due to sync" rather than "deleted due to user action"07:03
chriadamso that subsequent syncs don't handle those deletions in the wrong way and push them up to the server07:04
dcalisteI'll try to see what I can do not to see the deletion at a sync because deleted from server, not to see them as local eletion on next sync.07:04
dcalisteA trick would be to set the deletion date before the sync date.07:04
dcalisteThen, they will not appear in the deleted list on next sync.07:05
dcalisteOtherwise, put a flag in a comment.07:05
dcalisteInitially, I synced from a small window because I was doing test to solve once and for all the bug MER#183507:06
chriadamperhaps not comments07:11
chriadambut could use the flags field07:11
chriadamthere might be a valid flag we could set there (or define a new one)07:12
dcalisteflag field is scratched when deleted.07:12
chriadamI wonder why..07:13
dcalisteIt's to save space I guess in the database, but anyway deleted events are kept ad vitam eternam, so it's useless I think.07:14
dcalisteI would like to change mkcal in that sense a bit in medium term I think.07:14
dcalisteHaving a way to mark event as deleted, without removing entries in custom properties and recursive…07:15
dcalisteAnd another way to purge the databse when the deleted event has been deleted upstream also.07:15
dcalisteLike that, we can remove this comment hack and we can reduce the database size when event are deleted.07:16
chriadamright, that sounds like a better direction than what exists currently07:16
dcalisteYeh, but medium term, because it's a change in API and a good amount of changes in sqlitestorage from mkcal.07:17
dcalisteI'm still in the brainstorm part of what API should I need and from which objects.07:17
dcalisteI think, from API point of view a purgeIncidence() from extendedstorage.cpp (implemented properly in sqlitestorage) would suffice.07:18
dcalisteBut it's still early stage thinking.07:18
chriadamsure.  maybe pvuorela has some thoughts there also07:19
pvuorelaoh hi. i'm maybe missing a bit context here :)07:21
dcalisteYeh, I can come with a proposal on paper later in mkcal. I'll add pvuorela in the box.07:21
dcalistepvuorela: hello ! It's about deleted event in mkcal.07:21
dcalisteThey are kept in database, except their properties like custom flags, or recursive…07:22
dcalisteI would like to change this and have two possibilities:07:23
dcaliste- one, mark an event as (locally) deleted, without touching its properties. These events are then listable by the existing deletedIncidences().07:23
dcaliste- two, purge events from database completely (including properties), usefull when events have been deleted upstream and we want to get rid off them.07:27
dcalisteCurrently _all_ events (without their properties, but with the full set of the components table) are kept forever.07:28
dcalistepvuorela: that's the medium term reflexion I have on mkcal evolution…07:29
pvuorelahm, could make sense. it seems a bit weird that you cannot (?) remove events so that there's no kind of trace left.07:31
pvuorelakcalcore upstream wouldn't have anything related?07:32
pvuorela(away for a short while->)07:34
dcalisteNo, I think it's related to the storage, so it's mkcal only, but I'm not 100% convinced, because kcalcore is providing API for storage (but no impl)…07:34
dcalisteOnce again, I need to investigate this further indeed.07:34
chriadamthat's my understanding too, but I could be wrong07:34
chriadamyep.  anyway, can discuss further later on I guess :-)07:35
dcalistechriadam: indeed, I need to continue to think about it before arriving with a consistent proposal.07:36
dcalistepvuorela: if you look at sqlitestorage.h where you have the SQL commands for the database, you can see that there is no DELETE FROM components entry.07:36
dcalisteThe delete components macro is indeed doing an update on the components table.07:37
dcalistechriadam: besides, I've pushed an additional commit on top of caldav !4507:37
dcalisteThis commit is removing code related to test for possible modification due to etag update.07:38
dcalisteNow that last modified date can be changed with mkcal being updated to
chriadamah right - jpetrell agreed that bree could test that PR#45 so I will ask her to spend some time on that this week07:39
chriadamif she has availability for that one07:39
dcalisteStill need to be tested further but simple tests of locally adding an event and sync it (thus add etag) does not result of this event being seen as modified on next sync.07:39
dcalisteThanks for testing this internally also.07:39
chriadamhopefully we get that one in before next meeting (if you think it's ready, and our testing doesn't show any issues)?07:40
dcalisteI need to test now if I modify a recursive event, or add an exception if everything is working expectedly.07:40
dcalistechriadam: MR!45 will conflict with the direction one, but I can rebase easily after one of them will be accepted.07:42
chriadamcool, thanks very much07:42
dcalistepvuorela: see my named comment previously, mainly I'm starting this brainstorm about deleted events in mkcal and I'll come back to you when I have a paper proposal for API changes and we can continue to discuss there. Today was mainly to share these very early thoughts about this topic.07:43
dcalistepvuorela: by the way, if you have time at one moment to look at https://git.sailfishos.org/mer-core/python3/merge_requests/11 (it's reenabling GPLv3 readline in a separatepackage  for Python), that would be great, thanks in advance.07:46
pvuoreladcaliste: alright. for kcalcore was pondering about calendar.h having deleteEvent() and deletedEvent().07:46
dcalistepvuorela: yeh, but calendar is just a representation of the data, then the storage can map this as they want and it seems to me that sqlitestorage decided to just mark the event as deleted (so the extendedstorage::deletedIncidences()) can work.07:47
pvuoreladcaliste: sure. on python i'm not following the pip part. only commenting out some lines for packaged files, but no configuration changes?07:48
dcalisteBut when marked as deleted, the custom properties are deleted and thus the event listed by deletedIncidences() have only partial information…07:48
dcalistepvuorela: yeh, strange, but on SDK it didn't compile out of the box (imho). Pip has been disabled in a previous commit but files are still packaged !07:49
dcalistechriadam: didn't have time this week to go further for the email folder sync MRs…07:50
pvuoreladcaliste: hm, builds in obs.07:50
dcalistepvuorela: yeh, that's why I call it strange, I cannot figure out where these files come from in OBS.07:51
pvuorelayea, strange but think that needs to be figured out before merging. for pip i think we should be having a subpackage for that.07:53
dcalistepvuorela: I can remove the pip stuff which is strange for the moment I agree. Particularly, now that I look again at it, indeed pip has been removed earlier (adding --without-ensurepip) but it has been reenabled after because this without flag is not here anymore.07:56
dcalisteSo I propose to remove the pip commit from the MR and I'll retry to see why it's not compiling out of the box in SDK and propose later another PR if something is missing (maybe in the build require).07:57
pvuorelasounds good07:58
dcalistechriadam: I think that's all for mkcal/buteo for this week, thanks for the meeting.07:59
dcalistepvuorela: great thanks.07:59
chriadamdcaliste: thanks very much for your effort as always07:59
chriadamif there's any particular MRs which need further attention, please don't hesitate to ping me08:00
chriadamI will ask Bree to test the two caldav PRs this week if possible08:00
chriadamother than that, have a great week!08:00
dcalisteThanks chriadam, have a nice week too.08:01
dcalistepvuorela: MR!11 in python3 updated (removed pip commit and MR message). I'm investigating the SDK building issue.08:02
pvuoreladcaliste: nice, thanks!08:03
dcalistepvuorela: the ensurepip issue is that pip is installed in SDK already and the ensurepip module is doing nothing in that case. See MR: https://git.sailfishos.org/mer-core/python3/merge_requests/1209:42
dcalisteSee for instance this (closed) issue https://github.com/pypa/pip/issues/635509:43
dcalisteSome users would like to be sure that ensurepip is looking only in the ROOTDIR when --root is given (like for us when building package), but some say that it's inconsistent with Python module discovery…09:44
dcalisteI'm applying one of the user proposal in related https://github.com/pypa/pip/issues/3063 to use the --ignore-installed option of pip.09:44
dcalisteBut this option is not available in ensurepip, so there is a patch in the MR to add this option to ensurepip and apss it in the Makefile.09:45
*** vals_ is now known as tango_10:30
adisbladisAnyone who happens to have a copy of Sailfish for the Gemini?13:23
r0kk3rzTheKit: ^^13:52
TheKitadisbladis, you mean flashable image? you can flash image from Jolla and OTA update it to
adisbladisTheKit: Flashable image. I can only download
adisbladisI'm trying to work around https://together.jolla.com/question/212150/sailfishx-gemini-version-31011-no-sound/13:57
adisbladisTheKit: I guess some older version would also be fine as long as I can OTA from that14:12
*** frinring_ is now known as frinring17:58

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!