Thursday, 2018-10-18

*** frinring_ is now known as frinring02:27
flav0rdudes i am confused about MMS settings !04:33
rydarethey can be confusing04:36
flav0rwell get this04:50
flav0ri am trying to setup on this provider04:50
flav0rthey are a MVNO with tmobile04:50
flav0rlik eon android settings they have all this stuff about mcc and mnc04:51
flav0rand apn type04:51
flav0rthats for MMS04:52
flav0rthen for data they have a port number04:52
flav0rwhere can all that stuff got in sailfish04:52
flav0ralso iphone apn is different04:52
flav0rmaddening !04:52
dragonchaserwhat are you folks using for searching/storing/listening to podcast on sailfish?06:40
PeperJohnnyI'm using gpodder06:40
PeperJohnnyHowever a new app appeared in the store podqast or something which looked interesting06:41
PeperJohnnyBecause gpodder isn't supported as of 201606:41
DylanVanAsscheThere's a new podcast app apparently, just released in the Store and on Openrepos.net06:55
dragonchasergpodder seems to do the trick for now06:56
PeperJohnnyIt has no sync to the though06:57
PeperJohnnyIn case you wanted to use it06:57
mvogtchriadam: test07:02
chriadammvogt: pong07:03
chriadamdcaliste: ping?07:03
dcalisteHello chriadam.07:04
dcalisteI'm correcting the issue with headers.07:04
mvogtdcaliste: hi07:04
dcalisteI've download the Qt script and checking what is wrong: appears that I didn't put the Qt copyright. My bad :(07:04
chriadamno problem, I've similarly missed it several times in my own contributions :-D07:05
chriadamCI helps find such07:05
chriadamanyway, regarding the "empty body, attachment-only email" did you have something to discuss with mvogt ?07:05
dcalisteYes indeed.07:05
dcalisteHello mvogt, thanks for sharing a bit of your time.07:06
mvogtno problem :)07:06
dcalisteWe discussed a bit this issue a long time ago, for disposition: in-line content:image mails.07:06
dcalisteWe moved then the QMailMessageContentDisposition stuff from QMailMessagePart to PartContainer, so the QMailMessage can have access to these informations.07:07
mvogtis that correctly handled?  a non-multipart message with the body containing an image, and disposition set to inline?07:08
dcalisteSee 2.10, it is valid.07:08
dcalisteSo, yes the case image only with disposition in-line is properly handled now.07:08
dcalisteBut, of course his cousin case is also appearing: disposition: attachment and content application/pdf07:09
dcalisteAnd this is not working because attachment is supposed to be in a MessagePart API-wise.07:09
mvogtI'm not sure that one makes sense.. the body is attached to what? :)07:09
mvogtso the intention is that only an attachment is displayed, and no message content?07:10
mvogtthe part data is the body, right?07:10
dcalisteExact see the mail structure of…07:10
flav0ri just made my own podcast app07:11
dcalisteThis mail is properly handled by QMF and is consistent:07:11
dcaliste- no part,07:11
dcaliste- one body,07:11
dcaliste- with the content of the PDF file.07:11
dcaliste- reporting disposition attachment and even the filename.07:11
dcalistemail.findAttachmentLocations() is returning an empty list.07:12
dcalisteand (mail.status() & QMailMessageMetaData::HasAttachments) is false.07:12
dcalisteWhich is expected since the attachlment finder code is looking into parts.07:13
dcalisteBut this mail has no part.07:13
dcalisteThe attachment is the body itself.07:13
mvogtyeah :)  Can we use a predefined token to indicate body location?07:13
dcalisteThat was my first thought, yes.07:13
dcalisteLocation indice = 0 is not used.07:13
mvogtI guess the HasAttachments logic can additionally test the disposition of the message itself07:13
dcalisteparts are numbered from 1.07:13
dcalisteBut doing this will break other parts of the API.07:14
mvogtwhy is that?  I would have thought only the part-finding code would interpret the part location ids?07:15
dcalisteMost of the code user of QMF will expect QMailMessage.partAt(Location) to work (i.e. returning a massagePart), which will not be the case if we authorize Location.indice = 0.07:16
mvogtyeah, that's nasty07:18
dcalisteYes, I don't see any solution without touching the API…07:19
dcalisteMy other thought was:07:19
dcalisteadding new API for attachment handling, findAttachmentLocations() is duplicated into findAttachmentIds() and there is a QMailMessage.attachment(id) that returns either LongString data or file path.07:21
dcalisteAnd we propagate this new API into QMF user code.07:21
dcalisteAnd we deprecate findAttachmentLocations().07:22
mvogtIs LongString currently exposed via the API?07:22
dcalisteNot sure, maybe it's just QByteArrays.07:23
mvogtI hope it isn't :-)07:23
dcalisteI'm looking again how nemo-qml-plugin-email is requesting download…07:24
mvogtI think the partAt() functions are the problem - if they were able to fail gracefully, it would be ok to have a valid Location that did not yield a part, perhaps with an additional isBody() test07:25
dcalistenemo-qml-plugin-email is using QMailMessagePart.writeBodyTo(folderPath).07:26
dcalisteLet summarize the issues we have:07:27
dcaliste- findAttachmentLocations() returning part only,07:28
dcaliste- QMailMessage.partAt(Location) returning MessagePart only,07:28
dcaliste- all QMF user code using the assumption that attachment are QMailMessagePart.07:28
dcalistePossible solutions:07:29
dcaliste- make Location being also possible to point to the QMailMessage itself when multipart is none.07:30
dcaliste- create a special QMailMessagePart that encapsulate a multipart/none QMailMessage so the partAt() semantic is preserved.07:31
dcalisteI know we're dealing here with 0.1% of existing emails, maybe even less, but well it's itching that QMF cannot treat tehm properly.07:32
mvogtanother option is to make a higher-level message content identifier, which is capable of indicating a part or the body, and have findAttachments return a list of those07:35
mvogtbut I think the synthetic QMMPart approach is probably the best, if it can be done07:35
dcalisteWhat do you mean synthetic Part? Returning a Part wrapper around the QMailMessage for Location.indice = 0?07:36
mvogtyes, having an instance of QMailMessagePart that is the body of the message07:37
dcalisteI think it can be done in the QMailMessage.partAt(Location) with a private member that is created on the fly and returned if indice = 0 and QMailMessage/multipart = none.07:37
dcalisteBut I find it a bit hackish.07:38
mvogtyes, very :)07:38
dcalisteBut better, we can "officialy" store the body as a QMailMessagePart when the type is multipart/none.07:39
dcalisteThen partAt() is not building wanything on the fly, but simply returning the private memeber.07:39
mvogtI would have thought that there were many places making the assumption that multipart == None means that there are no parts07:40
dcalistemvogt, I agree, particularly in contentmanager for instance when storing the mail on disk and separating the body of parts from the main structure.07:41
dcalisteI think this would still be valid since the public API of QMailMessage is not changed and .body() will still give it properly, it's just that internally in private impl we store the body has a part.07:42
dcalisteThan it is valid to return it from partAt(location) with a special location = 0.07:43
mvogtI think, if you do it that way, you would still need a special hidden instance of QMailMessagePart in the message, because we would not want the body content to also show up in a listing of the parts (of which there are theoretically zero) - you don't want to add a part and find that it is in the second location07:45
mvogtso I07:45
mvogtam not sure that this solution is really cleaner than the previous one :)07:46
dcalisteYes, I agree that there is a QMailMessagePart in the private impl of QMailMessage. The cleaner aspect is that it is not populated when calling partAt() but when parsing the mail to store body content when multipart is none.07:48
dcalisteI agree it's not much cleaner ;)07:48
dcalisteSure this private part should not be listed when asking for parts.07:48
mvogtI think this case is rare enough that we should probably handle it in the client code - I think we should do the following:07:51
mvogt- allow Part::Location to refer the body07:51
mvogt- allow the Location to be tested to see if it points to a part or the body07:51
mvogt - deprecate the partAt() functions07:52
mvogt- replace them with locatePart() functions that return a pointer and can fail07:52
mvogtthen the client will have to test if the locations of the attachments are parts or the body, and handle them separately07:53
mvogtfindAttachments will also need a new argument to indicate that the client is opk with receiving a body location in the results07:54
mvogt'is ok', I mean07:55
dcalistemvogt: ok thank you for the guidance. I'm going to look at nemo-qml-plugin-email to see which changes are necessary to handle findAttachment with "extended" locations.07:56
mvogtif we do those things, have we actually saved the client any effort at all?07:56
dcalisteCurrently, nemo-qml-plugin-email is using a lot the fact that attachments are located is a QMailMessagePart(), so I think the changes will be important. But IMHO it's not that bad, since we're dealing with message parts up to the very last moment when saving them to disk.07:58
mvogtdcaliste: actually, what if we just change the HasAttachments logic?  Then in this special case, HasAttachments == true and findAttachments returns an empty list.  The client can then know that the body is an attachment...07:59
dcalistemvogt: yes that can be a clean solution also from QMF pov.07:59
mvogtthere is no case where the attachments list contains both the body and one or more parts07:59
mvogtBut probably better to add a new flag for BodyIsAttachment, since the client will have to understand the implications anyway08:00
dcalisteAs I said two lines ago, it may be sane to look into nemo-qml-plugin-email and to break the assumption that attachment are message parts but switching earlier that they are files to be saved somewhere.08:01
dcalisteFor the client, they have the disposition=attachment for the QMailMessage that indicates this.08:01
dcalisteWe're checking already disposition=in-line in the client anyway.08:02
dcalisteOk, so action: modify the hasAttachment status bit logic to return true in the right cases.08:03
mvogtI wonder how other clients handle the body-as-attachment mails?08:03
dcalisteKeep findAttachment as it is.08:03
dcalisteclaw-mail display nothing and provides an icon for attachment.08:03
mvogtseems reasonable :)08:03
dcalisteI think it's the right way. A placeholder for body would be nicer.08:04
dcalistefollowing action: look at nemo-qml-plugin-email and change the assumption attachment == part.08:04
mvogtthanks.  have you got somebody to review n-q-p-e changes?08:05
dcalistepvuorela is usualy giving quick feedback on nemo-qml-plugin-email.08:06
mvogtcool :)08:06
dcalistemvogt thanks for the discussion, I was really undecided on which way to follow, didn't see any solution without drawbacks. Sometime in the coming future I will come with PRs in QMF and nemo-qml-plugin-email about what we decided.08:07
dcaliste…or further questions if something was unexpected!08:08
mvogtyes, the unexpected should always be expected08:08
dcalistemvogt, chriadam: I'm trying to push the corrected headers for the signature PR, but I'm getting this message: "! [remote rejected] refs/for/master -> refs/for/master (no new changes)"08:13
dcalisteWhile current patch 10 is git revision 1a7cc27ff6164ea003b685902535390ccc0f805408:13
dcalisteAnd my local branch is git revision 67a1230910f2a8952e288278c6b9fb5332c6abd808:13
dcalistemvogt, chriadam: oups sorry, forgot HEAD in push cli…08:14
dcalistemy bad.08:14
mvogtcool, I wasn't going to come up with the answer :)08:14
dcalisteSo I've just pushed patch 11 with revised headers that are passing (locally) the script.08:15
mvogtI will approve it08:15
dcalistemvogt, thank you.08:16
mvogtwas it just the one PR, not the other one in that build?08:16
dcalistePR didn't pass neither, but it was complaining about missing headers in crypto files, which are not part of this PR…08:17
dcalistemvogt, thank you for review, hopefully I didn't leave any other mistakes and yet another CI roundtrip won't be necessary ;)08:24
mvogtno worries, CI roundtrips are not a big issue for qmf :)08:25
mvogtit's not like breaking qtbase08:25
dcalisteSure, this is another (several others) level(s) ! But leaving mistakes like that requires human maintainance which is a bit time wasting from your side…08:26
flav0rdoing paid work really interferes with time for my projects08:46
Mister_Magisterflav0r: captain obvious08:47
flav0rthats GENERAL obvious to you dude08:47
flav0rwell it was interesting enough08:48
flav0rslapped together a little SMS based set of menu items for people to get access to some arbitrary data08:49
IngvixI hope sf3 would come as soon as possible and would contain alien-dalvik update for at least lollipop. It's so frustrating not being able use my banking app which recently started requiring at least lollipop08:50
mvogtflav0r: what is SMS, above?08:50
flav0rtext messages dude08:51
flav0ror are you asking something different?08:51
mvogtoh, are you talking about the podcast app?08:51
mvogtoh :)08:51
flav0rsomething else08:51
flav0rjust something i happened to be working on thats paid work08:52
flav0ri was lamenting how it got in the way of projects08:52
flav0rthen i expanded a bit as to what i was doing specificall08:52
mvogtI see, I was confused about how you used SMS in your podcast app :)08:52
flav0rquite OFF TOPIC !08:52
flav0rwell the kicker is08:52
flav0rthe podcast app is for iOS !08:53
mvogtquite off-topic, indeed :)08:53
flav0ri am contemplating porting a couple apps i made to SF08:53
flav0rone is a youtube app08:53
flav0rand the other stuff is just for internal stuff08:54
flav0ri looked at some youtube app in the SF store08:54
flav0rseemed SLOW !08:54
flav0rwas kinda lame how apple pulled that type of app from their app store08:54
flav0rhad to make my own08:55
mvogtthat is the nature of apple :)08:55
flav0ra couple years back i got one of their latest apple tv’s through a developer program for $108:55
flav0rthey were trying to get people to develop apps for it08:56
flav0rso i made a porn aggregator app08:56
leszekhmm... don't we have enough youtube apps?08:56
flav0rthen i found out .. no porn in theirt store !08:56
flav0rwell i dunno man08:56
leszekmake a youporn app that is missing for SFOS :P08:56
flav0rall the youtube apps i see are kinda weak08:56
flav0rand completely do not fit my usage pattern08:56
leszekview youtube videos works fine for e08:57
flav0rmine is fast man … no latency … no ads obviously (as do the others)08:57
Ingvixare you speaking of ios or sfos?08:57
flav0ralso its lightweight .. small feature set08:57
flav0rstrip out audio too to save space (if desired)08:58
flav0roh i currently have that on ios08:58
flav0rSTILL off topic !08:58
flav0rmaybe ill port it08:58
flav0rive only done a little bit of work with Qt08:58
IngvixWell I like the new µtube on sfos. First one that can actually play streams08:59
leszekflav0r: lls vplayer does not show ads xD08:59
flav0ryah i imagine if anyone goes to the troubl of making an app they certainly wouldnt show ads08:59
leszekalso you can download youtube videos09:00
flav0ryah that was my main thing so i could play stuff offline09:00
leszekIngvix: you mean live streams?09:00
Ingvixyes, those09:00
leszekoh nice. Did not know09:00
Ingvixno browser on sfos could play them either09:01
flav0rwhats the “best” browser for sfos ?09:01
leszekyeah I think it is blocked for mobile browsers09:01
flav0ri had a quick look on the store and didnt see any !09:01
leszekflav0r: try my webcat browser. Though install the qtwebkit 5.212 update from my openrepos dev account aswell otherwise you will have an too old qtwebkit version to enjoy browsing09:02
IngvixI don't really know. I use the default one. Couldn't really find a real reason to keep using others. Some things work better on the default browser and some on webkit based browsers like webcat.09:03
flav0ranyone got any recommendations for a good phone that is highly compatible with SFOS that has a 4” screen09:03
leszekthough yeah webcats problem is not the featureset I assume but more the rendering engine. QtWebkit 5.212 hasn't been updated since january and I fear it is abandoned again09:03
leszekso basically the biggest flaw right now in sailfishos is the lack of a modern webbrowser09:04
flav0ryoud imagine that might be the first thing to get right09:05
leszekQt 5.9 with QtWebEngine might solve this though it needs a browser using it aswell :)09:05
leszekflav0r: it was much better in the beginning of SFOS. Though now we are stuck for years with Gecko 38 as engine in the sailfish-browser09:06
flav0rwell i guess i cant complain too much09:06
leszekthat has security issues all over the place and does not support modern web technologies09:06
flav0rmy first thought is … “how hard can it be” ??09:06
flav0rbecause im so used to stuff on ios working so nice09:06
flav0rbut then apple DOES have a lot of money to throw at the problem09:06
leszekyeah apparently it is hard. The 45 something update never arrived and that version is outdated now aswell09:07
flav0rin the next few months i am planning to retrofit an old brick style phone so it will work on modern networks09:10
flav0rmaybe that will solve my problems with web browsers09:10
flav0rsince it wont have a screen !09:10
*** phdeswer_ is now known as phdeswer12:19
dcalisteHello abranson, do you have some time to discuss a tracker issue ?14:44
abransondcaliste: sure what's up?14:44
dcalistetracker-store is creating a huge (900MB) meta.db-wal file.14:45
dcalisteLooking at the Internet, the wal file is supposed to be a transaction file.14:45
dcalisteSo not growing oversized.14:45
dcalisteWhat can I look at to give proper bug report ?14:46
dcalisteabranson, I will have to go but here is the status:15:08
abransondcaliste: i'd give it some time. in recent upgrades we've seen the wal file get quite large after an update. it should shrink back down once it's finished doing any reorganizing it seemed to be doing.15:14
dcalisteabranson, sorry I need to go: but yes that's my experience. And it was blocked on huge size because no more space was available.15:15
abransondcaliste: ok, speak later about it. sorry got distracted.15:16
abransona tracker reset might help then15:16
abransonand having a bit more space free in /home15:16
dcalisteThen, I clean a bit my home, giving some more hundreds of MB and the tracker-store then was able to finish his doing and now the wal file disappeared.15:16
dcalistethanks abranson for listening. Sorry not having a proper bug report at hand.15:17
dragonchaserhmm my xperia just froze completly when opening the browser...17:49
dragonchaserhow can I reboot it now?17:49
dragonchaserholding down the power button does not work17:49
DylanVanAsscheHold the power button for > 10-15 seconds17:49
dragonchasertried that... nothing happens17:49
DylanVanAsscheTry much longer then...17:50
dragonchasergetting cramps in my thumb atm17:50
dragonchaser40s still nothing17:50
DylanVanAsscheHmmm f it's not helping after 60s, no idea then :(17:51
Ingvixvolume button up + power button does a hard reset17:51
DylanVanAsscheIngvix: Nice to know, thanks!17:51
dragonchaserIngvix: thx17:52
dragonchaserthat does the trick17:52
tortoisedocwhat might this mean in obs :
tortoisedocI thought build jobs were self-contained?18:19
r0kk3rzthey are18:38
*** feodoran is now known as Guest1485823:19
*** feodoran_ is now known as feodoran23:19

Generated by 2.17.1 by Marius Gedminas - find it at!