mozilla :: #devtools

7 Sep 2017
09:11solejdescottes: ochameau I managed to get a build without the addon-sdk! Someone in the office helped me navigate the file
09:12solegoing to run a try and see how many thigns break
09:12firebot ASSIGNED, Perform a last "sweep" and make sure devtools works without the SDK
09:12soleBUT devtools opens!
09:12jdescottessole: yay!
09:12soleand no errors are printed or anything
09:13jdescottessole: you should stop stealing my commit messages though
09:13jdescottesI have "yolo" copyrighted
09:13solejdescottes: do you use yolo too?
09:13soleI felt so excited it was the first time I used yolo
09:13soleI thought the occasion required it
09:14jdescottesah so toolkit.mozbuild
09:14jdescottesgood to know
09:16soleThat was very hidden and mischievous!
09:26pbrogreat news if devtools opens and appears to be running!
09:27soleI switched between panels and that used to be the easier way to trigger addon-sdk 'failures' when removing their files, but now it was clean! a good sign
09:27solealso run a few tests locally and seemed successful but I want to use my computer for other things so pushed to try
09:28solehaha the title is try: #yolo
09:56jdescottessole: looks like your push is missing the changeset that touches toolkit.mozbuild
09:57solejdescottes: agh
09:57solejdescottes: I thought this would be enough
09:58jdescottessole: ah you're right I'm not used to having changes embedded in the changeset that contains the try syntax
09:58solejdescottes: rpl is also telling me to comment out other things in a dm
09:59jdescottesok don't know why the build is failing then :)
09:59soleApparently the addon-sdk tests!
09:59solejdescottes: don't worry, will figure this :)
10:48sole <- they want Java to GO FASTER
10:48solemoving to github too
11:58solejdescottes: Oh! I tried and it seems that you can use mach try when using cinnabar too, without having to embed the try syntax on the commit message
11:58soleSo much better
11:58soleIt&#39;s sole learns all the things week
12:07solejdescottes: zer0 : ochameau is it safe to remove these methods? should we have a follow up bug? (see mentions to &#39;addon-sdk&#39; in this file)
12:09ochameausole: I think so, I referenced that in bug 1397452
12:09firebot NEW, Remove all SDK compatibility code from devtools
12:10soleah thank you ochameau
12:11soleI had in fact already seen this bug! But I didn&#39;t connect the two in my mind...
12:19PorkepixHi, little question. Do we agree multi-process mode is e10s, and is on on Nightly for now a very long time? I wanted to see something in the adaptative view and got the following message: ; this is French, here is the English version of the string:
12:20PorkepixI got this on a Nightly. What could be wrong with detection? Worth filing a bug I guess?
12:22DuClareIs multiprocess actually disabled?
12:24DuClareAh hmm
12:26PorkepixDuClare: Well, looks like it was even if it was a Nightly. No idea since when nor why. I did a reset to few values with e10s in about:config and restart, I&#39;ll see what&#39;s happening
12:26solejdescottes: rpl PROGRESS! I removed the lines Luca mentioned (and more) to avoid jetpack mentions and now it seems my build isn&#39;t busted.
12:27soleOddly enough the build did fully build locally without removing jetpack mentions (without artifacts)
12:28soleurgh I don&#39;t know how if the build isn&#39;t busted because I am using artifacts there now, I&#39;ll request a build without them
12:29DuClarePorkepix: Most likely blocker is addons
12:29PorkepixDuClare: Well, extensions.e10sBlockedByAddons is setting itself to true while I only have two installed addons, and those are Webextensions. (uBlock and uMatrix). Aren&#39;t WebExtensions supposed to be e10s-compliant by default?
12:30DuClareI thought so, I don&#39;t know what&#39;s the logic behind it
12:31DuClareubo does indeed block it on my end too.
12:36PorkepixDuClare: Looks like without disabling any addons, after activating the multiprocess in the GUI preferences, and then reset all e10s lines in about:config and restarting it&#39;s now working as intended
12:36PorkepixI guess at some point there was a mess in Nightly, and profile wasn&#39;t back to a clean state through updates :/
12:36PorkepixWhich might lead to many nightly users actually having e10s disabled I guess
12:41rplsole: yeah, unfortunately the test infrastructure is very likely to run some stuff that we don&#39;t regularly run locally (I&#39;ll keep an eye on your last pushes to try, but feel free to ping me at any time if I can be helpful, even just as a second pair of eyes)
12:45solerpl: Waiting on the last try, but the one with artifacts seems promising...
12:46DuClarePorkepix: There&#39;s telemetry e10s_stataus but I don&#39;t know if I trust it .. 0 samples for disabled for addons :|
12:47PorkepixDuClare: No idea what is this
12:50rplsole: great! once the full build is also happy, I&#39;m wondering if it could be a good idea to run a &quot;./mach try ... -u all&quot;, just to double-check that the other &quot;unit tests&quot; suites are happy too
12:51solerpl: Ah I can do that too, I wanted to start with making sure devtools worked :D
12:51solerpl: but it&#39;s a good idea too!
12:52nchevobbeI&#39;m getting &quot;abort: push creates new remote head 2b757f0fd0a7 with bookmark&quot; when pushing to try
12:52nchevobbeanyone knows what could be wrong in my repo ?
12:52rplsole: absolutely, and it makes completely sense to start from a single test suite until that one passes without issues
12:53solerpl: glad to confirm I am in the right path! Thank you :D
12:53soleIt&#39;s not every day that one deletes entire folders from Firefox...
12:55rplhehe, definitely, especially one that is there from a good amount of time and its references have been already spread everywhere :-)
12:57soleIt gives you a sort of trepidation when you write git rm -r addon-sdk 8-)
13:03solerpl: build NOT busted! \o/
13:03soleThere&#39;s a failing devtools test but who knows why
13:04solehas previous failures etc
13:05rplsole: great! \o/
13:13solerpl: and the build is pretty much green!
13:14solegreen builds
13:14solehaha so tacky
13:14soleI&#39;m going to try all the tests and see how it turns into OrangeFEst
13:15rplhaha, &quot;OrangeFest&quot;... I&#39;m going to steal this joke ;-)
13:15solerpl: like this
13:31solezer0: Honza: jdescottes meeting?
13:31jdescottessole: can&#39;t join today sorry
13:31zer0sole, i&#39;m having a small issue with vidyo, I&#39;m trying
13:31Honzasole: joining
13:33pbrosole: interested in joining if not too late. Which room is it?
13:33pbronevermind, I&#39;ve got a conflict
13:35solepbro: ha it;s the devtools one
13:40solepbro: but we&#39;re gone. Ha ha!
13:46zer0jdescottes: if you&#39;re too busy to work on Bug 1395741, I can take it.
13:46firebot NEW, Stop using SDK loader in adb-helper
14:08smallfoot-If the server replies with JSON I can see that in the devtools in a tree structure. But there is no way to copy the JSON to a clipboard?
14:09smallfoot-like if you do a GET you can do it
14:09smallfoot-but if you do a POST you can&#39;t do it in devtools
14:20nchevobbesmallfoot-: is the network item disappearing ?
14:20nchevobbesmallfoot-: it might be that you need to enable persistent logs
14:23smallfoot-nchevobbe, no I see the POST response, I click on the &quot;Response&quot; tab, it shows the JSON tree. But there is no way to copy the tree to clipboard
14:25nchevobbesmallfoot-: oh okay, I see
14:27jdescotteszer0: hi! re Bug 1395741 I have it on my todo for later today, but if you have some free time for it right now, you can steal it
14:28firebot NEW, Stop using SDK loader in adb-helper
14:30pbrosmallfoot-: that&#39;s a good point. It would be good to file a bug for it. Do you know where? Honza is the person who can guide you on this. He has done work on the netmonitor lately
14:31smallfoot-I see
14:32Honzasmallfoot-: Can you use &quot;Copy POST Data&quot; action?
14:32smallfoot-also when devtools is docked to the side, then Inspector, Console, Debugger, Style Editor, Performance, Memory and Storage all show okay, but not Network, the requests take upp all vertical space, leaving no room for the Header, Cookies, Params, Response tabs
14:33smallfoot-Honza, I want the response returned by the server, not the data sent by the client
14:34nchevobbesmallfoot-: in the response tab, you can copy the &quot;response payload&quot; content
14:34Honzasmallfoot-: There is &quot;Copy Response&quot; for POST requests too.
14:34Honzasmallfoot-: Does it work for you?
14:34smallfoot-nchevobbe, by select with mouse? That copies the rendered tree, not the JSON data
14:35nchevobbesmallfoot-: the response payload is a text area, so you can cmd + a to select all
14:35nchevobbeand then copy
14:35smallfoot-Honza, yes I found that now, the &quot;Copy Response&quot;, however it works on those 200 GET, but copies the wrong data on this 400 POST. I don&#39;t know if thats a bug
14:36smallfoot-nchevobbe, it is not a text area. The payload from the server is JSON. So it is a data tree.
14:37Honzasmallfoot-: Sounds like a bug. Can you please report it and include some steps to reproduce? An having a test page would help a lot.
14:37nchevobbesmallfoot-: are you looking what&#39;s below &quot;Response payload&quot; toggle header thing ?
14:37nchevobbesmallfoot-: can you post a screenshot here please, maybe i&#39;m misunderstanding you
14:37smallfoot-nchevobbe, no, I look at the &quot;Response&quot; tab which is the 4th tab.
14:38zer0jdescottes: no, that&#39;s fine! If you&#39;re going to do today there is no rush about it. We just mentioned during the stand up. If, by any means, you got too busy to work on it today, feel free to assign to me!
14:39nchevobbesmallfoot-: yes, me too :) look at
14:41smallfoot-nchevobbe, my Firefox looks different from yours. My only have &quot;> JSON&quot; not &quot;Response payload&quot;
14:43jdescotteszer0: ok will let you know in case :)
14:48solerpl: it&#39;s not too bad!
14:49solerpl: Not quite the OrangeFest I was expecting
14:51rplyeah, it is only a small glass of orange juice :-), this one seems to be one of those references to the SDK tests that are still there:
14:54jdescottesso sole, are we no-sdk yet?
15:04solejdescottes: I think *we* are but the rest of the tree isn&#39;t very happy about letting go of it!
15:05solerpl: well I don&#39;t mean to fix jetpack issues, so I was kind of ignoring these failures
15:05jdescottessole: \o/ (ok sad about rest of the tree but at least we didn&#39;t miss anything else on devtools side apparently)
15:06solePerhaps John-Galt can have a look and be like &quot;no sole, you totally missed this other thing that is not obvious and will make devtools fail&quot; maybe
15:06soleI don&#39;t know how to tell the tree not to run those mochitests, I thought I already did!
15:31jdescottessole: wrong bug number I think :)
15:32solejdescottes: what?
15:32jdescottessole: just saw a weird update on
15:32firebotBug 1397287 FIXED, [nosdk] Stop using addon-sdk OS in devtools/shared/system.js
15:32solejdescottes: oh yeah
15:32rplsole: I recall that there are two places that references the jetpack tests for the build infrastructure, and one was off of the m-c tree (and iirc it was used to run jetpack tests in non e10s mode) , I&#39;ll take a look asap
15:32solejdescottes: this was megayolo
15:33solerpl: so I put all the commits on a review for you and John-Galt to have a look
15:33soleI think I failed at &quot;all the commits on a review&quot;
15:33soleit&#39;s also here
15:33firebot ASSIGNED, Perform a last &quot;sweep&quot; and make sure devtools works without the SDK
15:37andymonce the devtools are done, i&#39;m thinking a &quot;hey we&#39;ve done it, we are about to completely delete the SDK, speak now or you are in trouble&quot; email might be nice :)
15:52soleandym: I don&#39;t want to be the person that deletes the SDK!
15:52andymhehe, i&#39;ll send the email and let John-Galt do it
15:52soleI wanted more for your team to give me a reassuring check
15:52soleandym: mind if I send the email?
15:52soleI was half way
15:52andymof course not, go for it
15:53solethank you! that way I can give links to my try submission etc
16:09soleandym: John-Galt: rpl: sent email!
16:10andymto where?
16:10andymah there
16:11soleandym: to your... email
16:49HonzaHow come I don&#39;t have Firefox build (binaries) for this try push?
16:54glHonza: I took a look, but also don&#39;t know why. I was curious from just looking at the patch since we also added scalar logging for the grid inspector that we needed new methods in telemetry.js
16:55Honzagl: yeah, I need to build Firefox to see the new scalar and I was hoping try can do it for me...
16:56glHonza: ya, that is one of the painpoints with adding telemetry
16:56gli added my probes here
16:56firebotBug 1373483 FIXED, Add telemetry for the grid inspector
16:56glbut i am curious about the new methods and if i needed to update my telemetry calls
17:00tromeyjryans: fitzgen: if a source map has two entries for a given location ... (1) which one should be used, and (2) is this a producer bug?
17:00tromeyI&#39;m inclined to think it must be
17:00fitzgentromey: inlining?
17:00fitzgenwhich direction?
17:00glHonza: I am mostly asking because I think we might actually need those methods and everything might actually be done with this.telemetry.log() and .toolOpened()
17:00glmight actually don&#39;t need*
17:01tromeyfitzgen: for (line 2,column 2) of the generated file, the map has entries to two different locations
17:01tromeyfitzgen: in this case it&#39;s from sass
17:01Honzagl: yes, sounds promising, in a meeting now
17:01fitzgentromey: sounds bogus to me
17:02tromeythanks fitzgen
17:03fitzgentromey: if it were the other way: multiple generated locations for the same source location, then that seems reasonable (could be explained by inlining for example)
17:03tromeymaybe source maps should have a monotonicity requirement
17:04jryanstromey: yeah, seems like a bad map to me
17:07tromeythanks jryans
17:19John-Galtsole: Oh, good, I was planning to `rm -rf addon-sdk` today and see what broke. Thank you for beating me to it :)
17:54tromeynode-sass sure makes it a pain to file a bug
18:10jdescottesJohn-Galt: maybe we should move the devtools copy of the loader we recently landed to a shared folder ?
18:11John-Galtjdescottes: We could, but that doesn&#39;t actually need the module loader. That copy of sinon is browserified, anyway. Other test code just loads sinon as a JSM
18:12jdescottesah ok!
21:59TitanNanoWhen debugging this: (just open devtools and type one key into the input) the execution order of promises gets messed up. The promise callback is executed synchronously instead of as asynchronously, as it should and normally does.
8 Sep 2017
No messages
Last message: 14 days and 3 hours ago