Thursday, 2018-11-08

*** frinring_ is now known as frinring02:57
dcalisteHello chriadam_, how are you ? Thanks for staging the correction in QMF.08:02
chriadam_dcaliste: doing ok, better than the last two days (neck issues).  sorry for not being available on Tuesday.08:03
chriadam_no problem, thank you for spotting that and fixing it :-)08:03
dcalisteNo problem, I was testing the full stack before leaving for vacation and notice that sent email was not acting properly any more. So many packages are involved that I didn't notice the regression.08:04
dcalisteI'm now testing the packaging of latest QMF in mer-core, see Compilation is almost finished. I'll then copy and install packages on device and test.08:07
chriadam_sounds good :-)08:08
dcalisteAbout the threading issues in nemo-qml-plugin-email, the deep-copy looks not so "deep" after all. The attachment are created as LongString(filename) which do not map the attchment content into memory as long as the content is not needed.08:09
dcalisteSo first deep copy (before running the thread) is light.08:09
dcalisteThe second deep copy (returning from thread after signature process) will have the attchment content into memory though.08:10
dcalisteBut this content is necessary for the signature process anyway.08:10
chriadam_well, which attachment content?08:14
chriadam_just the new signature data (as a new attachment)?  or all the existing attachments (e.g. if images are there, could be expensive)08:15
dcalisteThe signature is done for attchment also, to guarantee that attchments are pristine.08:15
dcalisteWell, in fact we can sign just the text part if we want, but the common behaviour is to sign all parts.08:15
chriadam_but it shouldn't need to _modify_ the attachments, should it?  it just calculates a signature for that data?08:15
dcalisteIncluding attchments like images.08:16
chriadam_so a deep copy shouldn't be required in that case, since nothing is being modified?  or maybe I'm misunderstanding08:16
dcalisteYes, the attchments are not modified.08:16
dcalisteWhat is modified is the structure of the mail…08:16
dcalisteYou come with {multipart/mixed 1. body 2. image1 3. image2} and should go out with {multipart/signed 1.{multipart/mixed …} 2. signature data}.08:17
chriadam_I see08:17
dcalisteSo the QMailMessage is indeed in/out in the process.08:18
chriadam_I still wonder if that will cause a deep copy of the actual attachment data, or just the QMailMessage object08:18
chriadam_if it's the latter, the cost is negligible08:18
chriadam__sorry, my internet has been flaky08:20
chriadam__dcaliste: it's fine in any case.  let's not get too concerned about it until it proves a problem in practice08:21
dcalisteI'm looking into longstring.cpp in QMF to see how the deep copy affects mapped files.08:21
dcalisteIndeed, the copy operator will create a new private structure which will setup the filename without actually copying the data.08:25
dcalistechriadam__: From my understanding, deep copying QMailMessage with attchments should not copy the actual content of attchments even when already mapped to memory, but just copy the filenames.08:26
dcalisteWhich looks great for our case.08:26
chriadam__yep, perfect08:26
chriadam__thanks for checking that one08:27
dcalisteSo thank you for pointing me the possible threading issue. Now verification and singnature processes should be properly independant of main thread behaviour.08:28
dcalisteThis will be particularly relevant with S/MIME which is much slower than OpenPGP.08:28
chriadam__thanks for fixing the signal connection race issue in secrets too!08:31
chriadam__(had a comment there regarding the commit message, if possible to fix that one)08:31
chriadam__but that was a very good catch08:31
dcalisteI've corrected the commit message and added the bug id.08:31
dcalisteWell, I didn't know about QtConcurrent, so when you suggested it for nemo-qml-plugin-email, I read the doc and look at how it was used in secrets.08:32
dcalistechriadam__: ok, mer-core packaging of latest QMF is working as expected when signing and verifying OpenPGP emails, tested on device with the full stack in email app.08:40
chriadam__somehow I managed to miss the doc about signal connection race ;-)  thanks for pointing that one out08:40
chriadam__nice one08:40
* chriadam__ heads home - will merge the secrets PR tomorrow08:45
chriadam__let me know if there's anything else which is blocked waiting for my action :-)08:45
dcalistechriadam__: sure thanks. Have a nice evening.08:45
chriadam__you too, thanks again08:45
dcalistechriadam__, you or pvuorela should approve if you think it's fine like that.08:46
chriadam__pvuorela_: ^^ I can check that one tomorrow also.  need to test that jolla-email still works without any changes I guess / existing accounts etc.08:47
dcalistechriadam__: yes, we can continue this discussion tomorrow.08:48
chriadam__thanks :-)  gnight08:49
pvuorela_dcaliste: yea, appears ok to me, but i haven't been looking at the full set of changes that much.08:51
dcalistepvuorela_: thanks for the previous MR, cleaning up a bit the account patch. I've rebased mine on this and I'm now using git formatted patch.08:51
pvuorela_good good.08:52
pvuorela_hey, i also did some small adjustment to your eigen packaging and removed the eigen2 parts from git history:
pvuorela_need to bug people to create mer repo etc.08:53
dcaliste;) great !08:54
dcalisteI'll go back to work on Calligra more often when PGP stuff are in.08:54
dcalisteBut I'll give a look indeed to your eigen3 stuff. When it's officially in, I'll update the calligra-bundle spec file.08:55
dcalisteBesides, I fully agree that switching to git format patch should be nicer there also…08:55
dcalisteI'll follow also your advice to use #fidef instead of comments to remove unsupported parts from Calligra sources.08:56
pvuorela_good :)08:56
martinst11Hi all. I'm currently writing a sailfish-app and having some trouble (I'm too stupid for QML...). What I would like to have as an initial page is a Page that looks like a dialog. But where I can open a dialog by swiping to the left (adding a new filter to a search-query) and swipe right for moving forward (triggering the search with all filters). Somehow I can't figure out how to do that. A dialog as an initial page seems not10:34
r0kk3rzyeah you need to have a backing page for a dialog10:47
leszekmartinst11: never saw a dialog as initial page. You need a page object as "initial page" :)10:47
martinst11It should only "look like" a Dialog. But I maybe could push a Dialog-object on the stack in front of the initial page. And then re-add a new one after the current dialog closed10:49
martinst11Hm, there doesn't seem to be a clean way to do this (there is no insert-function for the pagestack). OK, maybe I have to resort to a button for opening the dialog (would have been nice without one though)11:01
kimmolisailfish-communi does something like this, it starts showing a dialog where it asks which servers to connect11:16
martinst11Had a quick look into that. I'm afraid what they are doing is a bit out of my league ;)11:31
Mister_Magistersooo can we block calls or this super advanced option didnt arrived to sfos yet13:06
Mister_Magisterdid they just… updated aliendalvik to 8.1?16:34
leszekMister_Magister: plan to update AD on XA2 to 8.116:43
leszekit won't be backported to X or earlier devices as it needs restructioring16:44
* Mister_Magister just watching video16:44
Mister_Magisterkinda sucks16:44
Mister_Magisterthey could just go for anbox16:44
leszekthat would mean start at level 0 again. But yeah open and free software I also would prefer for this. However we don't know how much they paid for AD16:45
Mister_Magisterbut free trial is nice16:45
Mister_Magistermy friend with xperia x compact will be able to update sfos16:45
Mister_Magisterleszek: i just sent sami a congrats SMS :P17:02
* Mister_Magister wonders if he still has this number its finland one so i suppose17:07
leszekpushAttached is broken for me. It does nothing when I pop from a page that had an attachedPage17:17
Mister_Magisterleszek: whatcha mean17:20
Mister_Magisterpushattached stays attached only when you push further17:21
leszekMister_Magister: this does not work when return from the video player page:
Mister_Magisterleszek: pushAttached stays attached only when you push further iirc17:22
Mister_Magistertest behaviour in components qt517:23
leszekMister_Magister: I don't see an error in my code17:23
Mister_Magisterleszek: just check behaviour17:23
Mister_Magisterif its supposed to stay17:23
Mister_Magisteror whatcha mean17:23
Mister_Magisterleszek: you mean its not pushed?17:25
leszekthats what I mean17:26
Mister_Magisterdo debug like webdev17:26
leszekit isn't pushed and it should be pushed17:26
Mister_Magisterput console.log next to it17:26
Mister_Magistercheck if its even running17:26
leszekit is running17:26
leszekit is jumping into the if17:26
Mister_Magisterwell shit17:27
leszekyeah I fear it has something to do that I pop a page that has its own attachedPage17:28
Mister_MagisterWelcome to JavaScript world17:29
Mister_Magisterleszek: i would do super simple app with one page and pushing page like that to just check if its supposed to work17:30
leszekpushAttached in general is working17:30
leszekand also in webcat this exact function is working17:31
leszekthe only difference is that I am now popping from a page which has an attached page17:31
Mister_Magisterleszek: i mean app for this behaviour17:31
Mister_Magisterpushing attached page when going back from page with attached page17:31
Mister_Magistersimple app to test it thats what i would du17:32
Mister_Magisterbut thats just me17:32
leszekmy lazyness is trying to fix it somehow with strange workarounds17:39
leszekoh god damn it found the error. Mister_Magister actually I am just stupid as fuck :P17:51
Mister_Magisterleszek: hmm?17:52
* Mister_Magister no he isnt updatig every port to 3.0 nao17:52
leszekthat gets called and removes my pushAttached page17:54
Mister_Magisteryou just pop before leave17:54
Mister_Magisterthats rather obvious thing to do but i would check if it werks without doing it17:54
Mister_Magisterbut das just me17:55
leszekMister_Magister: it works after removing it. Though a bit strange. It shows the popped Page from the videoPlayer :P17:56
Mister_Magisterdunno whatcha mean but okay17:56
leszekI am confused17:59
martonmikloshi all!21:28
martonmiklosI am working on this QML plugin21:28
martonmiklosbut I cannot get it use in QML any way21:28
martonmiklosI got Cannot assign to non-existent property "vCardSerializer"21:28
martonmikloswhen I declare an item from it with vCardSerializer {         id: id_vCardSerializer     }21:29
martonmiklosin QML21:29
martonmiklosthe plugin's registerTypes and initializeEngine got called21:29
martonmiklosso I think my library file is in the correct place21:29
martonmiklosany idea would be welcome21:30
jef91I'm using the pre release of SailfishX on my Xperia and since upgrading my audio and video are desynced on android apps like YouTube and Hulu23:23

Generated by 2.17.1 by Marius Gedminas - find it at!