Gaia/SMS/Scrum/2.1S8

From MozillaWiki
< Gaia‎ | SMS‎ | Scrum
Jump to: navigation, search

List of bugs

SMS issues handled by the SMS subteam (blocks the sprint bug 1090150)

Bugzilla link

Full Query
ID Assigned to Summary Blocking b2g Feature-b2g Whiteboard Resolution
1080820 Steve Chung [:steveck] [Messages][RTL] Sticky header does not take the whole width --- No cf_feature-b2g [p=1] FIXED
1080828 Oleg Zasypkin [:azasypkin] [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed --- No cf_feature-b2g [p=2] FIXED
1081207 Oleg Zasypkin [:azasypkin] [Messages] We never revoke the blob URL for the attachments 2.2+ No cf_feature-b2g [p=1] FIXED
1089920 Julien Wajsberg [:julienw] [Messages] Investigate and fix the gaia-header in Messages app (~50ms) 2.2+ No cf_feature-b2g [p(2.2S5)=1][p(2.2S3)=1][p(2.1S9)=1][p(2.1S8)=3] FIXED

4 Total; 4 Open (100%); 0 Resolved (0%); 0 Verified (0%);


Remaining points and burndown chart

google chart api url for Sprint 2.1S8

Burndown chart
Remaining points
Start 8
Day 2 7
Day 3 7
Day 4 7
Day 5 6
Day 6 6
Day 7 5
Day 8 3
Day 9 3
Day 10
End


SMS issues handled by the SMS subteam outside of the sprint (contains whiteboard "sms-sprint-2.1S8")

Full Query
ID Assigned to Summary Blocking b2g Feature-b2g Whiteboard Resolution
1079824 Steve Chung [:steveck] [Messages] Draft saved from activity is duplicated 2.2+ No cf_feature-b2g [sms-sprint-2.1S8][p=1] FIXED
1090821 Julien Wajsberg [:julienw] [Messages][Drafts] Can't save message draft with attachment 2.2+ No cf_feature-b2g [sms-sprint-2.1S8] FIXED
1090945 Julien Wajsberg [:julienw] [Messages] icon for an "email" recipient is too big on Flame --- No cf_feature-b2g [sms-sprint-2.1S8] FIXED
1092021 Steve Chung [:steveck] [Messages] Report panel visual refresh follow up for word truncate and l10n issue --- No cf_feature-b2g [sms-sprint-2.1S8] FIXED
1092112 Oleg Zasypkin [:azasypkin] [Messages] Waiting screen never disappears if user confirms thread deletion when there isn't any selected thread 2.0M+ No cf_feature-b2g [sms-sprint-2.1S8] FIXED

5 Total; 5 Open (100%); 0 Resolved (0%); 0 Verified (0%);


All SMS issues tracked for this sprint (target milestone)

Bugzilla link

Full Query
ID Assigned to Summary Blocking b2g Feature b2g Resolution
1030672 Salvador de la Puente González [:salva] [Sora][HOMO][o2 Germany][o2 Bug 56785]Regional] DUT doesn't send MMS to email accounts --- --- FIXED
1058526 Vishnu Teja [:ythej] Poor grammar in MMS error message --- --- FIXED
1080820 Steve Chung [:steveck] [Messages][RTL] Sticky header does not take the whole width --- --- FIXED
1080828 Oleg Zasypkin [:azasypkin] [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed --- --- FIXED
1081207 Oleg Zasypkin [:azasypkin] [Messages] We never revoke the blob URL for the attachments 2.2+ --- FIXED
1083198 Augustin Trancart [:autra] [B2G][SMS] Remove fallback for old gecko version in MessageManager.getMessages --- --- FIXED
1089198 Wilson Page [:wilsonpage] Tap on message report header, the contextual menu "…" will be shown - --- FIXED
1090150 SMS sprint 2.1S8 --- --- WONTFIX
1092112 Oleg Zasypkin [:azasypkin] [Messages] Waiting screen never disappears if user confirms thread deletion when there isn't any selected thread 2.0M+ --- FIXED

9 Total; 9 Open (100%); 0 Resolved (0%); 0 Verified (0%);


Sprint planning

Minutes are on a separate page.

Daily meetings

Day 2: 29th October

Steve

  • bug 1074732 - [Messages] Create a mixin to handle the "select" UI model
    • Landed in master
  • bug 1085764 - [flameKK][v2.1]Be able to input Simplified-Chinese in SMS subject after reaching maxium length
    • WIP patch attached, but some bugs discovered. One is the suggestion will override the chars that not belongs to the typed word in suggestion, another bug is the contenteditable element might receive the input event while removing the element. Will describe more on the bug.
  • bug 1090043 - [B2G][Message] MMS attachments missed in SMIL could not be displayed
    • Create a patch for it and hope it could solve the unsupported format issue as well.

Today:

  • Waiting feedbacks from SMIL and subject length limit bugs.
  • Select all message/threads method refactoring.
  • Code reviews

Julien

  • bug 1071514 (flickering while adding an attachment, blocker)
    • landed !
  • bug 1076729 (scroll the thread view to the end, even when loading MMS)
    • I did a very small patch for this. We can see that the window is flashed when a MMS is appended, there can be different reasons for this. yes, there is a reflow, but I think the reflow could be expensive because of some CSS we have (notably last-child...)
    • in review !
  • perf startup improvements (bug 1087329 and dependencies)
    • tested bug 1089145 (remove StickyHeader init from the launch path) but I didn't see visual improvement. But I noticed I had a bigger issue on Buri that may have hidden this, I'll retry soon now that it's fixed.
    • bigger issue is... your gaia-icons.ttf file packaged in the app needs to be the same than the one in your device (/system/fonts/hidden) otherwise gecko loads it later and we have a visible glitch + reflow.
  • sent a mail about drafts behavior to the team; I'd really like that we fix the small behavior glitches we have
  • others: scrum stuff, smart data stuff

Today:

  • will do reviews + needinfo
  • will send a mail to the layout ML
  • will send a mail to dev-gaia with a link to our demos
  • will work on the header reflow issue

Oleg

  • bug 1081207 - [Messages] We never revoke the blob URL for the attachments
    • Started to look into this, should have patch ready for review today (in progress)
    • Question: does anybody know a) when exactly object\blob url is disposed after it was revoked or it's up to GC? b) does revokeObjectURL wait for exact URL returned from createObjectURL or we can pass objectURL + mozSampleSize fragment and fragment will be ignored?
      (Julien) (a) as soon as it's revoked you can't use it anymore (means, if you set it, and revoke before gecko loaded it, it fails when gecko loads it later). But when the underlying blob is disposed (if no other references) is up to the GC (b) I don't really know but I'd go with the safest: use the exact same URL :)
      (Oleg) (a) Hmm, probably I'm missing something - tried with jsfiddle and opened blobURL image in a separate tab, then revokedURL and image is still can be retrieved - will check again (http://jsfiddle.net/b9k5jo59/6/) (b) - yep, currently use safe way :)
      (Julien) (a) you're doing something strange because when you load the blob url in a separate tab, you're not in the same document anymore. I don't know how this is supposed to work.
      (Oleg) :) Okay, will check with the same document, but it works cross tabs :) That's the way I used to understand whether gecko ignores fragment.
      (Julien) with http://jsfiddle.net/73fsv9ra/ I can't load it anymore, but I don't know if that was what you meant
      (Julien) about fragments: even if it works now, we're not sure whether it will still work in the future; so yeah, stay safe with this :) yep :)
      (Julien) let's follow-up on IRC about the blob url

Other:

  • Read through all updates from previous sprint;
  • Some time to dig into smart data spec/cost control app - lack of integration tests is a bit scaring, but good that we don't need to update old with these huge changes :)

Today:

Day 3: 30th October

Steve

  • bug 1085764 - [flameKK][v2.1]Be able to input Simplified-Chinese in SMS subject after reaching maxium length
    • Have some discussion with Rudy(keyboard owner) and seems we have 2 issues and need platform support. Maybe we can split the bug into 2 issue, one is for suggestion, and another is for Chinese IME(When length reach to max, input with in composing mode should be prohibit => but choosing word from suggestion will still exceed, it's belong to 1st issue).
      (Julien) if not the same solution for both issues, then yeah, maybe file 2 bugs
  • bug 1090043 - [B2G][Message] MMS attachments missed in SMIL could not be displayed
    • r+ and with nit addressed. wait for try server
  • bug 1091350 - [FFOS2.0][Woodduck][MMS]MS can't auto retrieve MMS if insert two SIMs and one sim close the data connection.
    • Still looking, trying to figure out why the mms service id is missing.
      (Julien) do you reproduce? I think having the placeholder means we have undefined or null for the values... should look at our uses of this string in v2.0 and see if there are cases where we don't pass the correct data.
      (Steve) I'm just checking code , but I need to find another validate sim for experiment first
  • Some work from system platform window manager
  • Reply some thought about the message draft

Today:

  • Land SMIL patch and mms service id missed in dual sim.
  • Select all message/threads method refactoring.
  • Code reviews

Julien

  • bug 1076729 (scroll the thread view to the end, even when loading MMS)
    • r+, needs to land, filed bug 1091441 to improve this more
  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday
  • mostly did reviews yesterday, they were piling up
  • did a small patch so that lazy loader supports promises (bug 1090966), this will be useful for us

Today:

  • will do reviews + needinfo
  • will send a mail to the layout ML
  • will send a mail to dev-gaia with a link to our demos
  • will work on the header reflow issue

Oleg

  • bug 1081207 - [Messages] We never revoke the blob URL for the attachments
    • Prepared patch and sent for review, still didn't cover case when attachment is removed manually with backspace key (in review).
      (Julien) do you have an idea to resolve this case?
      (Oleg) I was thinking only about tracking all attachments in Set\Map instead of Weak one and revoke all URLs on Compose.clear by iterating through attachments Set\Map
      (Julien) could be a simple solution, yep, given we always eventually clear.
      (Oleg) I remember we had a bug when we wanted to show confirmation before removing attachment with backspace - I thought initially that this code can be a good place to revoke URL, but looks like decided (per Omega) that confirmation is too strong
      (Julien) you're refering to bug 991835 right? yep, but he still wanted to change the behavior, but not sure how easy it is
      (Oleg) Yep, I meant that bug
      (Steve) Maybe ni? Jenny for this one again? I didn't prefer the confirmation either...
      (Julien) yep :)
  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Started working on it (in progress).
      (Julien) if you want to discuss ideas on IRC, I'm here :)
      (Oleg) Thanks!
  • bug 1090821 - [Messages][Drafts] Can't save message draft with attachment
    • Noticed that bug while working on "revoke URL" bug, per regression window looks like gaia patch caused it.
      (Julien) yep I'll have a look later today, but it's still strange.

Other:

  • Some small review requests

Today:

Day 4: 31st October

Steve

  • bug 1085764 - [flameKK][v2.1]Be able to input Simplified-Chinese in SMS subject after reaching maxium length
    • Fire a bug about the Chinese IME(bug 1091550), it looks more like a blocker
  • bug 1090043 - [B2G][Message] MMS attachments missed in SMIL could not be displayed
    • Landed in master
  • bug 1091350 - [FFOS2.0][Woodduck][MMS]MS can't auto retrieve MMS if insert two SIMs and one sim close the data connection.
    • The behavior is expected, it should close as invalid, but I could not figure out the note from reporter. And I could not reproduce it anyway...
      (Julien) could be a RIL issue from their side? If it happens again they'll likely fire a separate bug...
      (Steve) Not sure since I couldn't reproduce that, and Bevis had no idea either. That what I suggested on the bug.
  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • I doubt another popup dialog is controlled in system app and handles the general cellbroadcast message. Ni? reporter to see if she could get the non-emergency-alert message.
      (Julien) on a side note, maybe I should ask for a new component Emergency Alerts in Bugzilla? So that we could watch this component. What do you think?
      (Steve) Definitly, it would be much better
      (Julien) will ask today; also maybe add a new Firefox OS modules in https://wiki.mozilla.org/Modules/FirefoxOS Thanks!
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • Some bugs from report view refresh. Patch created.
      (Julien) thanks ! \o/
  • Some work from system platform window manager

Today:

  • Investigate cellbroadcast issue.
  • Select all message/threads method refactoring.
  • Code reviews

Julien

  • bug 1076729 (scroll the thread view to the end, even when loading MMS)
  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, I bricked my Buri I use to compare before/after... thanks Settings team :/
  • some more reviews
  • bug 1090821: investigated the IndexedDB issues we had, they're clearly my fault from bug 1071514
    • When saving the attachment as a draft to IndexedDB, the object has a _renderer property, that contains a DOM node.
      (Oleg) I think "priv" would help here? I mean putting such private props to weakmap
      (Julien) I originally did this but Steve thought that because this is used a lot we didn't want the (supposed) performance penalty. So now I simply added a toJSON() function to the Attachment.prototype that I use in Compose.getContent().
    • also, now that we don't display the resized image anymore, we need to increase the max_thumbnail_size so that all images have a thumbnail. Not sure of the appropriate size with moz-sample-size (and it's really dependant of width/height more than real size). I use 5M for now, Oleg told me it works ok in Buri.
    • patch coming today
      (Steve) We could attach multiple images at once, not sure if it's safe enough. May need more testing about this.
      (Julien) yep, because my buri is bricked right now, I can't really test. Maybe I can lower the flame's memory though.

Today:

  • will produce a patch for bug 1090821
  • will do reviews + needinfo
  • will send a mail to the layout ML
  • will send a mail to dev-gaia with a link to our demos
  • will work on the header reflow issue if I can revive the buris

Oleg

  • bug 1081207 - [Messages] We never revoke the blob URL for the attachments
    • Working on covering "delete attachment with backspace" case (in progress). Should be ready today.
  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Read through variety of RTL materials (this one is my favourite :) http://www.w3.org/International/articles/inline-bidi-markup/ - there is a bunch of cases covered including numbers) to better understand how the things should work;
    • I think I've covered all cases mentioned on the bug, just need to make it prettier and write unit tests where needed. I believe should be ready for review today (in progress).
  • bug 1079824 - [Messages] Draft saved from activity is duplicated
    • We're observing this issue again (only on master) :( Verified that SharedWorker patch isn't the culprit. Let's wait for the regression window first.
  • bug 1053952 - [Messages][Refactoring] Delete all messages at once when deleting threads
    • Was just asked to uplift this to v2.0M, to fix "bug 1087991 - [Woodduck][SMS][interaction]MS SMS function will always stay in the deleting sms interface unless plug out battery.". Hesitating a bit whether we need to uplift or just find simple fix for the mentioned issue. What do you guys think?
      (Julien) I think it's better to have a simple fix if you can find the root reason. If you can point the reason maybe someone from Device Team can handle it then?
      (Oleg) Ok, will try to spot what the real problem is.
      (Julien) I already had a quick look but didn't find the right cause :/
      (Steve) Could it be reproduced on 2.0?
      (Oleg) Will try to reproduce on 2.0, haven't tried yet
      (Steve) If it's not reproducible in our 2.0, we could ask help from device team.
      (Oleg) Makes sense, will check that, thanks!

Other:

  • Some small review requests

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.

Day 5: 3rd November

Steve

  • bug 1091350 - [FFOS2.0][Woodduck][MMS]MS can't auto retrieve MMS if insert two SIMs and one sim close the data connection.
    • Close as invalid first, although our partner still can't fully understand this use case
  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • Need to create a patch to filter out the CMAS message in system. Or maybe we should moving this part into Network alert in the future.
      (Julien) you mean handle all CB messages in network alerts?
      (Steve) yep , any idea about this?
      (Julien) I think I proposed that when we started working on network alerts but it was too much work. So I'm +1 ;)
      (Steve) Ya, we've talked about this at the beginning, not sure if we can make this decision by ourself, or need more discussion with system
      (Julien) it's really a refactoring/technical debt reduction; so yeah we need discussion with system, but also I think it takes too much time and we have more interesting things to do... :/
      (Steve) Agree...
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • In review.
  • bug 1080820 - [Messages][RTL] Sticky header does not take the whole width
    • Create a patch but it will delay the header show up time because of white background color. But moving this part into shared will make the calllog unable to see the header in missed call tab. Not sure if it's a potential bug in calllog, but I might take 3rd suggestion that copy the element background before refresh.
  • Some work from system platform window manager

Today:

  • Create a patch to filter out CMAS message in system.
  • Refine the sticky header patch.
  • Select all message/threads method refactoring.
  • Code reviews for attachment regression(Sorry :/) and deletion overlay.

Julien

  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, but since I finished with bug 1090821 (hopefully) and my Buri are working now, I can resume today.
  • bug 1090821: investigated the IndexedDB issues we had, they're clearly my fault from bug 1071514
    • mostly worked on this, thought of a simpler patch, asked for a review

Other:

  • took part to the bug squash last saturday and helped 2 contributors on simple patches.

Today:

  • will handle reviews bug 1090821
  • will work on the header reflow issue
  • will do reviews + needinfo
  • will send a mail to the layout ML

Oleg

  • bug 1081207 - [Messages] We never revoke the blob URL for the attachments
    • Got r+, landed (landed).
  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Found more cases that still needed to be covered, working on it, should be ready for review/feedback today. Probably we'll ask for feedback before fixing tests :) (in progress)
  • bug 1092112 - [Messages] Waiting screen never disappears if user confirms thread deletion when there isn't any selected thread
    • Prepared small patch for this issue (in review).
    • The underlying problem is that newly received message causes target thread item to re-render that unchecks thread checkbox, so that we don't have any thread to delete and waiting UI never disappears.
  • bug 1078384 - [MMS] Move MMS label to appear always at the end of the subject line
    • Gave feedback, still needs some work.
    • Noticed that once we add subject it's not focused, I thought we had this functionality earlier - will check.
      (Julien) look at the earlier UX specs for version 1.3 (I think that's where we got the subject support), the link is in the main wiki page. Or just ask Jenny ;)
      (Oleg) Okay, thanks!

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.

Day 6: 4th November

Steve

  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • Will give a patch today.
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • Got r+, will reply question later.
  • bug 1080820 - [Messages][RTL] Sticky header does not take the whole width
    • Create another patch that don't involve the call log and sticky header. It only applied css changes in message app and solve the initial display issue.
      (Julien) but it doesn't look very nice :/ why don't you like adding a class? Using an attribute selector here is very fragile (imagine we change the implementation).
      (Steve) Well, it's true that not robust enough, but having a class but only used in one app seems weird.
      (Julien) the class works as a marker; it's not an issue if another app does not use it. Maybe ask for double advice from Kevin? If he doesn't want a class then let's move with your solution.
      (Steve) Ok, will create a patch for class and request feedback from Kevin.
  • bug 1091960 - Gaia incorrectly checking for CMAS alerts settings
    • Have some discussion with platform, we will need to display 2 Presidential Alerts channel by default. And we should not have to deal with CMAS in CDMA case.
  • bug 1091511 - [MGSEI][Flame][v2.1][Message]The number/email or URL can't be tapped if there is full-width text in this message
    • Report as blocker originally, but it's not a regression since the root case is full-width char. Not sure if we need to handle Chinese word, will ni? Jenny later.
  • Some work from system platform window manager

Today:

  • Create a patch to filter out CMAS message in system.
  • Refine the sticky header patch.
  • Select all message/threads method refactoring.
  • Code reviews for attachment regression(Sorry :/) and deletion overlay.

Julien

  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, again.
  • bug 1090821: investigated the IndexedDB issues we had, they're clearly my fault from bug 1071514
    • fixed small nit, ready to land
  • reviewed contributor patches from Bug Squash, landed one of them
  • reviewed more patches from Oleg/Steve
  • Investigated the duplicate draft bug 1079824 with Ben Turner, and I think this is coming from the SMS app actually (but exposed by Ben's IndexedDB patch). I'll add logs today and try to understand the issue.

Today:

  • will work on the header reflow issue, first goal is to try Wilson's patch from bug 1074783
  • investigate bug 1079824 (likely not produce a patch though, just trying to find what the issue is)
  • will do reviews + needinfo
  • will send a mail to the layout ML

Oleg

  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Got first feedback! Will handle comments and reply review questions (in progress)
  • bug 1082067 - Receiving 2 SMS with prepaid SIM card breaks SMS app
    • Investigated this blocker, as far as I can tell it reproduces even in 2.0. At the first look the problem seems in "getSelf" and inner alert interaction eg. following code snippet repeated twice in any panel "navigator.mozApps.getSelf().onsuccess = () => alert(1);" will break app completely: navigation doesn't work (fb requests never returns, localization is broken too). Having setTimeout(..., 0) around alert resolves issue for me. Will try to have simpler test case and probably ask Alive about it (in progress).
      (Julien) normally, one alert is making all events ignored; so here I'd say that having 2 alerts in the same time (which should not be possible... "onsuccess" should likely not be called if we already have 1 alert) make it that it's not cleaned up correctly :/ likely a platform issue.
      (Julien) and if it happens in older versions, maybe ask for blocker status again
      (Julien) what happens with setTimeout around alert? Do we still have 2 alerts?
      (Oleg) Yep, still two alerts. Alerts are handled by system app, right?
      (Julien) yeah, but go throuh Gecko first; I think Alexandre Lissy knows this well, but likely alive too. Vivien too but I don't know how available he is these days.
      (Oleg) Ok, thanks for the pointers

Other:

  • Handled feedback requests from :salva, and small review request from Steve.

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.

Day 7: 5th November

Steve

  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • WIP given and got positive feedback, will complete test part today.
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • Oleg find some small issue about the patch(another invalid l10n and css selector), so refine the patch and request another review
  • bug 1080820 - [Messages][RTL] Sticky header does not take the whole width
    • Got r+ and landed
  • bug 1091511 - [MGSEI][Flame][v2.1][Message]The number/email or URL can't be tapped if there is full-width text in this message
    • It's interesting that Jenny thought we should deal the full-width text but partner didn't think it's important. But I think handling full-with punctuation marks is reasonable IMO
      (Julien) I think so, unless it's costly. Maybe we can wait for/request Gecko support for full ES6 unicode. (see https://mathiasbynens.be/notes/es6-unicode-regex ) "astral" symbols. Not sure if full-width chinese characters are astral symbols :)
      (Steve) Looks cool, but if we only need to handle limited marks, I think we don't really need full ES6 support?
      (Julien) yes, maybe we don't need this :) I mean, ES6 support is not mandatory to support this, it just makes things easier if we fall in these cases. But let's do it if this is requested by partner or if you think it's good :) (but in a next sprint :p)
  • Some discussion about copy/paste in message app, PM eagers to land this feature for 2.2, but they got some concern about message bubble copy. They book a meeting at non-Europe friendly time (3:AM). I can tell them to change the time if you are willing to join.
    (Julien) there are some bugs about it already, did you see them? I'm fine if you lead this topic :)
    (Steve) The main concern is they don't want the text selection cross bubble. They comes out some solutions but not really ideal. Will discuss this a little more on IRC
  • Some work from system platform window manager

Today:

  • Complete patch about filter out CMAS message in system.
  • Land report view refresh follow up patch.
  • Select all message/threads method refactoring.
  • Update the conclusion for message copy/paste
  • Code reviews for dialog transition regression.

Julien

  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, again.
  • bug 1090821: investigated the IndexedDB issues we had, they're clearly my fault from bug 1071514
    • landed
  • reviewed more patches from Oleg/Steve
  • Investigated the duplicate draft bug 1079824 with Ben Turner
    • found out root issue and provided some guidelines to fix it
  • Investigated a strange bug happening on a v1.3 phone sold in France, but it looks like it does not happen with a newer update from ZTE (bug 1092951 - most text is french ;) )
  • Investigated why the font weight looks incorrect in contacts prompt (bug 943165)
  • reviewed sticky header patch from Steve (bug 1080820)
    • favored the solution that involves adding a class in the sticky header page
  • reviewed bug 1091299 that is a regression from the dialogs' animations
    • we'll add pointer-events: none to document.body while the animation happens
    • I saw that merely adding a class to document.body, even if it does match no rule at all, produces a noticeable restyle operation that prevents the animation from running (at least in Buri with my data). But adding "document.body.style.pointerEvents = 'none'" does not have the issue (even if Etienne told me it should do a reflow). So looks like this specific reflow is not noticeable.
  • looked at bug 1091751 about the double CMAS alert; I still think we should uplift to v2.1 but maybe we'll need to change the patch to not filter out CDMA.
  • looked at bug 1093665: a message is not appended in the right position. Strange but likely because of the time issues we have on Flame.
  • looked at bug 1092840 (and bug 1092838 from same person) but this is in v1.3... but the SMS app is completely unusable for the user :/

Today:

  • will work on the header reflow issue, first goal is to try Wilson's patch from bug 1074783
  • will look at contenteditable issues in bug 1094000
  • will do reviews + needinfo
  • will send a mail to the layout ML (maybe)

Oleg

  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Fixed unit tests + some updates per review/feedback comments (in review)
    • Will file several bugs about RTL issues discussed in review.
  • bug 1082067 - Receiving 2 SMS with prepaid SIM card breaks SMS app
    • Investigated a bit more, looks like all issues I've noticed (broken localization, hanging of FB request) come from the fact that once two alerts are fired inside getSelf().onsuccess callback - setTimeout stops working :) Alive forwarded question to platform guys, they should be looking, in the meantime as per Alive - alert has other issues too, so it would be good to migrate to CustomDialog. I think it's a good idea to get rid of alert everywhere in our code.
    • Prepared WIP CustomDialog patch for current issue (in progress)
      (Julien) we have another similar bug where alert add "Messages" automatically and Jenny would like to remove it. It makes me really sad that we can't keep window.alert in a web application :/
      (Oleg) I don't really like alert because it's not automatically localizable and it's hard to have title and message body
      (Julien ) yeah, it has a lot of issues, but it 's so much easier :) And I don't like CustomDialog a lot because it inserts new DOM. And I think we wanted to get rid of it at one point. But maybe we don't want to anymore... But I got this from Vivien early in the project.
      (Oleg) "inserts new DOM" - mmm didn't get concern :)
      (Julien) as you probably noticed ;) we try to have the DOM ready at the start, and merely hide/show it. This was a good practice that I got when I joined the project. But maybe the practice changed.
      (Oleg) Got it, anyway if it was because of performance impact, I think Gecko should handle this in a good way, as a lot of modern web app use this as well as frameworks. But yeah, it's offtopic :)
      (Julien) I think our main perf issues are reflows and restyles, not DOM adding/removing (even if it's considered as "slow", I don't think it's a real issue for a handful of elements).
      (Oleg) Yep

Other:

  • Handled few review/feedback requests.

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.
  • Will create test case for RTL vs Ellipsis issue and file a bug

Day 8: 6th November

Steve

  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • Landed in master and request 2.1 approval.
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • Oleg find some small issue about the patch(another invalid l10n and css selector), so refine the patch and request another review
  • bug 1079824 - [Messages] Draft saved from activity is duplicated
    • Ongoing, need to add recipient reset in ThreadUi cleanField
  • bug 1092437 - Messages app should follow text selection pattern
    • Update the status for the bubble copy discussion
  • Some work from system platform window manager

Today:

  • Land report view refresh follow up patch.
  • Duplicated draft bug
  • Code reviews for dialog transition regression.

Julien

  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, again.
  • Investigated why the font weight looks incorrect in contacts prompt (bug 943165)
    • provided new screenshots for Fang, just needs a last change from contributor to land
  • reviewed bug 1091299 that is a regression from the dialogs' animations
    • we'll add pointer-events: none to document.body while the animation happens
    • I saw that merely adding a class to document.body, even if it does match no rule at all, produces a noticeable restyle operation that prevents the animation from running (at least in Buri with my data). But adding "document.body.style.pointerEvents = 'none'" does not have the issue (even if Etienne told me it should do a reflow). So looks like this specific reflow is not noticeable.
    • I discussed quickly with some layout peers but had no definite answer; I'd need to investigate more but I don't really have time. In the mean time another solution was provided for the bug that I'll need to review today.
  • bug 1094000: found the core issue with the cursor location, filed bug 1094633. If this bug is fixed we'll be able to remove all work around we have around keeping br tags in contenteditable ! yay!
    • the "editor" peers will likely fix another bug when they don't insert BR in some situations. But I really think inserting BR is a workaround :/ I think they need to do it just to not break backward compatibility for this under-specified feature :/
  • started looking at bug 918970 again (from a french contributor). It will be really cool and it's quite close to the end: when starting an activity for a number, showing the existing thread if it's existing
  • reviewed Oleg's patch in bug 1080828 for RTL

Today:

  • will work on the header reflow issue, first goal is to try Wilson's patch from bug 1074783
  • will do reviews + needinfo
  • will send a mail to the layout ML (maybe)

Oleg

  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Fixed last review nit, added desktop mocks with RTL names, got r+, buuut looks like bug 1094320 is blocker :( Currently it's noticeable only when we switch language, but if we don't deal with it now - we'll have other problems in the future.
      (Julien) I see you NI me on this, I'll have a look. Could be a good argument against the "all is webcomponents" trend - sometimes being generic just doesn't work.
      (Oleg) Yeah, it's not web component prime time yet, but in this particular case it sounds like they just don't want to do complex work :)
      (Julien) but I can understand this, and tht's the problem with generic code: it needs to accomodate all cases, making the code complex, even for the simple cases. Sometimes we need specific stuff. Maybe we can fork gaia-header and have gaia-complex-header that could build on it. I don't know, just throwing ideas :)
      (Oleg) Agree, if it's architectured correctly it should be "opened to extension" :)
      (Julien) yes. But that's not really the trend our web component team is taking. It's more "let's fix all the cases with one code". Anyway, I'll look at it :)
      (Oleg) Thanks!
  • bug 1094320 - It's impossible to have rich HTML markup inside heading elements (h1, h2, h3, h4) if they defined inside gaia-header component
    • Noticed, investigated, found a reason - innerHTML of all headings (up to h4) inside gaia-header is replaced with corresponding textContent - that is IMO a bug in gaia-header. Wilson doubts that he can ever fix it, so asked for workaround from our side.
    • I can only imagine using LTR/RTL marks, but it's hack and haven't checked it yet.
      (Julien) wanted to ask something but will look at it first and then I'll ask on IRC :)
  • bug 1082067 - Receiving 2 SMS with prepaid SIM card breaks SMS app
    • Looks like Kan-Ru can have quick and not complex fix at Gecko side, I think we'd better wait for him and don't do migration to CustomDialog yet - Gecko related bug is "bug 1094083 - [BrowserAPI] Handle nested-alert gracefully". Julien should be happy to leave alert :)
      (Julien) yeah, it means less code to maintain for us :p (see for example all the crap we need to do because we added transitions in custom dialogs)
      (Oleg) that's true :(
  • bug 1094131 - [Messages][RTL] Ellipsis truncates first part of the LTR contact name in Thread List
    • Created test case and confirmed with Ahmed that this is issue indeed, ni? Ehsan. I've heard that Ehsan is OOO for some weeks, do you who else can answer to Gecko LTR questions?
      (Julien) I don't know :/ it's maybe not that urgent, but maybe it can move out of his sight when he comes back. So just don't forget to ping him on IRC maybe 1 week after he comes back ;)
      (Oleg) Yeah, wanted to say the same :) Yep won't forget, also added as a dependency to our RTL meta bug

Other:

  • Handled review request from Steve, r+'d.

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.
  • Will check if LTR mark can help to workaround bug 1094320

Day 9: 7th November

Steve

  • bug 1091751 - Sending two kinds of message for cellbroadcast-received
    • Wait for QC 2.1 branching, this dependency seems not easy to maintain our own branches.
      (Julien) yep, I think I'd rather uplift it so that the burden to revert it is on their side :/ maybe we can ask if it's ok to do this.
  • bug 1092021 - [Messages] Report panel visual refresh follow up for word truncate and l10n issue.
    • Landed in master
  • bug 1079824 - [Messages] Draft saved from activity is duplicated
    • Create a tiny WIP patch that remove the visibility event listener.
      (Julien) sorry, haven't looked yet, but will for sure today
  • bug 1092437 - Messages app should follow text selection pattern
    • UX spec released, since contentEditable div could not limit the selection range(unlike the input/textArea, select all inside these element will limit the range only inside)
      (Julien) after all I wonder if this could be fixed at "selection feature" level? I don't know how it works.
      (Steve) "selection feature"? Gecko side once proposed user-select: element that could treat other element selection like input, but this idea was withdrawn...
      (Julien) I mean, how does "select all" works? why can't it select only inside contenteditable? Anyway the spec is also about the bubble, not about contenteditable, after all.
      (Steve) Because of "inputs" are all contenteditable div actually we have to switch user-select frequently, not only the bubble. For example, the compose will need to set user-select: none by default and only selectable when focus on.
      (Julien) oki!
  • Some work from system platform window manager

Today:

  • Duplicated draft bug
  • More copy/paste experiment
  • Start select all implementation.

Julien

  • perf startup improvements (bug 1087329 and dependencies)
    • nothing new from yesterday, again.
  • reviewed bug 1091299 that is a regression from the dialogs' animations
    • reviewed the new solution but does not seem very practical so I'd like to stick with pointer-events: none.
  • discussed about bug 1094076 with gecko developers for scroll/APZ/marionette
    • marionette should work just fine for "tap" but maybe it doesn't. Actually marionette should scroll automatically before tapping, but it looks like it doesn't work properly :/
  • looked at bug 1094320 (rich content in a gaia-header), have comments but not really solutions. We can talk on IRC about this.
    • (Oleg) Yep let's talk

Today:

  • will work on the header reflow issue, first goal is to try Wilson's patch from bug 1074783
  • will do reviews + needinfo
  • will send a mail to the layout ML (maybe)

Oleg

  • bug 1080828 - [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed
    • Landed, found one feftover :( (landed)
  • bug 1094822 - [Messages][RTL] Number of group thread participants in Thread panel heading should always be LTR
    • Prepared tiny patch (in review).
  • bug 1094320 - It's impossible to have rich HTML markup inside heading elements (h1, h2, h3, h4) if they defined inside gaia-header component
    • Tried to use LTR and RTL marks to workaround gaia-header bug, but wasn't able to get the same effect, looks like dir="auto" for the contact name (maybe both LTR and RTL of course) in group thread header is the only solution.
  • bug 1082067 - Receiving 2 SMS with prepaid SIM card breaks SMS app
    • Dependent Gecko fix is in inbound branch! Let's wait when it reaches PVT to verify.
    • Unfortunately as QA can't reproduce with real use case (message-0), probably once it's landed we'll ask reporter.
  • bug 1092112 - [Messages] Waiting screen never disappears if user confirms thread deletion when there isn't any selected thread
    • Was uplifted to v2.1 and v2.0m, do we need v2.0 approval request? Nobody expressed interest in it :)

Other:

  • Handled review request from Steve, r+'d and Salvador, f+ with some suggestions;
  • Prepared small fake-moz-activity patch for the contributor working in Firefox\Mulet: https://github.com/azasypkin/gaia/compare/bug-desktop-moz-activity.patch
  • Started digging through CostControl blockers - not much to what we can pick up. Started to create first Marionette JS test for the app (with mocks of course :(). Also was able to run their app in Mulet (again with the the same build script & mocks technique we use in Messages).
    (Julien) very great, I wanted to do this too, I think this will help us a lot :)

Today:

  • Will handle review/feedback requests;
  • Will work on review comments and assigned bugs.
  • Will continue with CostControl Marionette JS patch and some desktop-only fixes - it's actually helps to understand how it works and what it needs :)

Day 10: 10th November

Demos

bug 1080820: [Messages][RTL] Sticky header does not take the whole width

bug 1080828: [Messages][RTL] International phone numbers and groupmms threads are not correctly displayed

bug 943165: Make the header for the contact prompt right

Thanks to Munto (Dorian) our contributor!

Retrospective

Previous sprints' actions

From 2.1S3:

  • Look into VM with gaia dev environment
    Nothing more done in this sprint

From 2.1S4:

From 2.1S5:

  • in planning for sprint N, we pick some items we want to do in sprint N+1 to ask early questions to UX and Designers (esp refresh bugs, but also bugs with UX changes)
  • we'll add screenshots of panels in Wiki, with the various cases for these panels, to help analyzing change impacts and not forgetting things
    Oleg started this: https://wiki.mozilla.org/Gaia/SMS/Current_App_State
  • ask UX about WIP status for v2.2, so that developers can give ideas and feedbacks
    UX is still not ready yet

From 2.1S6:

  • Steve tries to arrange local meetings in the morning ;)

From 2.1S7:

  • Julien should file bugs early so that he can leave the ideas alone :) "file bug and forget for some time" (Oleg's expression)
    (Julien) did it at least once ;) (bug 1079824)

What was good in the last sprint

  • New component network-alerts created, Clear dependency that we don't have to mix all the bugs into message component.
  • pretty much a good sprint, I think we moved forward quite well with pending features

What was bad in the last sprint

  • I'm not very happy with the gaia-header bug, I couldn't make it move as much as I wanted, and it's the 3 left points (bug 1089920)
    (Steve) Should we list some possible way for further investigation? Like font fit issue and nested issue
    (Julien) font-fit was used in v2.0 too... The "nested" issue is only in v2.2 because we don't use nested markup in header earlier. Further investigation is profiling v2.1 vs v2.0, I think. I can try to do this this afternoon (since it's the end of the sprint ;) ).
    (Julien) maybe my fault here is that I didn't report earlier that I couldn't make it.

Any questions

  • not sure how we'll work once we start Smart Data. Maybe Steve should take the owner role?
    (Steve) You mean message?
    (Julien) yes :) for Messages
    (Julien) not urgent yet, I still don't know if you'll be part of Smart Data team or not...
    (Steve) You are right, let's wait until we get clear situation. Maybe message app wont have new feature in 2.2 than.
    (Julien) from what I understand, only RTL is in v2.2. But I can expect also performance improvements. So yeah maybe v2.2 can be very quiet for Messages :)

Actions for next sprint

Nothing planned.