1383830 - Can no longer drag and drop into tab content

Status: VERIFIED FIXED

(Core :: DOM: Core & HTML, defect, P1)

Description

[Markus Stange [:mstange]]

I think this is a recent regression.

Dragging files from Finder into a content tab no longer works. But dragging them into a parent process tab (like about:support) works.

Examples:
 1. Dragging an HTML file in order to load that file.
 2. Dragging a file on top of a file <input>
 3. Passing a file to web content, e.g. for loading a local profile file on https://perf-html.io/

Comment 1

[Andrew Overholt [:overholt]]

Seems like it works on Windows. Michael, have you seen anything that could have broken this? Maybe Stephen has an idea?

Comment 2

[Nika Layzell [:nika] (ni? for response)]

I haven't touched any D&D code in months, so I have no idea what might have broken it. I can try to look into it once I get back from PTO if someone hasn't gotten to it by then.

Comment 3

[Stephen A Pohl [:spohl]]

Ran mozregression. Looks like bug 1363601 introduced this. I've been able to confirm by building locally from that cset.

 9:35.40 INFO: Narrowed inbound regression window from [217ad633, e7a3ac60] (3 revisions) to [217ad633, 1044cf05] (2 revisions) (~1 steps left)
 9:35.40 INFO: No more inbound revisions, bisection finished.
 9:35.40 INFO: Last good revision: 217ad633fbf7ea7b334c1da31c96248424f1d1dd
 9:35.40 INFO: First bad revision: 1044cf0518e931651ea51b6b407838807972703c
 9:35.40 INFO: Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=217ad633fbf7ea7b334c1da31c96248424f1d1dd&tochange=1044cf0518e931651ea51b6b407838807972703c

Comment 4

[Gabor Krizsanits (INACTIVE)]

(In reply to Stephen A Pohl [:spohl] from comment #3)
> Ran mozregression. Looks like bug 1363601 introduced this. I've been able to
> confirm by building locally from that cset.
> 
>  9:35.40 INFO: Narrowed inbound regression window from [217ad633, e7a3ac60]
> (3 revisions) to [217ad633, 1044cf05] (2 revisions) (~1 steps left)
>  9:35.40 INFO: No more inbound revisions, bisection finished.
>  9:35.40 INFO: Last good revision: 217ad633fbf7ea7b334c1da31c96248424f1d1dd
>  9:35.40 INFO: First bad revision: 1044cf0518e931651ea51b6b407838807972703c
>  9:35.40 INFO: Pushlog:
> https://hg.mozilla.org/integration/mozilla-inbound/
> pushloghtml?fromchange=217ad633fbf7ea7b334c1da31c96248424f1d1dd&tochange=1044
> cf0518e931651ea51b6b407838807972703c

That patch is a change in a comment on nightly. It is a no-op patch on the nightly channel, so I think something might went wrong in this mozregression run...

Comment 5

[Andrew McCreight [:mccr8]]

Backing out bug 1363601 for me locally didn't fix the issue for me. I did notice that sometimes the drag and drop DOES work, even though mostly it doesn't. Maybe that was an issue when doing a regression?

It probably isn't useful, but in my debug build I saw this when it failed (sometimes with more WARNING):
[Child 59924] WARNING: NS_ENSURE_TRUE(dragSession) failed: file /Users/amccreight/mc/dom/base/nsContentUtils.cpp, line 5934
[Child 59924] WARNING: NS_ENSURE_TRUE(dragSession) failed: file /Users/amccreight/mc/dom/base/nsContentUtils.cpp, line 5934
[Child 59924] WARNING: NS_ENSURE_TRUE(dragSession) failed: file /Users/amccreight/mc/dom/base/nsContentUtils.cpp, line 5934

When it succeeded the log the log was:
[Child 59924] WARNING: NS_ENSURE_TRUE(dragSession) failed: file /Users/amccreight/mc/dom/base/nsContentUtils.cpp, line 5934
[Parent 59923] WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80470002: file /Users/amccreight/mc/netwerk/base/nsFileStreams.cpp, line 74
[Parent 59923] WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80470002: file /Users/amccreight/mc/netwerk/base/nsFileStreams.cpp, line 74

Comment 6

[Andrew McCreight [:mccr8]]

I'll run another mozregression on it and see if I get another range.

Comment 7

[Stephen A Pohl [:spohl]]

(In reply to Gabor Krizsanits [:krizsa :gabor] from comment #4)
> (In reply to Stephen A Pohl [:spohl] from comment #3)
> > Ran mozregression. Looks like bug 1363601 introduced this. I've been able to
> > confirm by building locally from that cset.
> > 
> >  9:35.40 INFO: Narrowed inbound regression window from [217ad633, e7a3ac60]
> > (3 revisions) to [217ad633, 1044cf05] (2 revisions) (~1 steps left)
> >  9:35.40 INFO: No more inbound revisions, bisection finished.
> >  9:35.40 INFO: Last good revision: 217ad633fbf7ea7b334c1da31c96248424f1d1dd
> >  9:35.40 INFO: First bad revision: 1044cf0518e931651ea51b6b407838807972703c
> >  9:35.40 INFO: Pushlog:
> > https://hg.mozilla.org/integration/mozilla-inbound/
> > pushloghtml?fromchange=217ad633fbf7ea7b334c1da31c96248424f1d1dd&tochange=1044
> > cf0518e931651ea51b6b407838807972703c
> 
> That patch is a change in a comment on nightly. It is a no-op patch on the
> nightly channel, so I think something might went wrong in this mozregression
> run...

It was a strange cset to settle on for sure. I just ran mozregression again and it settled on a different range[1], pointing at bug 1351148 as the possible culprit. I rebuilt locally before and after the patches in that bug and it seems to confirm that these patches introduced the issue.

[1] https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=bb33237155278066423e2fdc3de2785eb44a75e7&tochange=0ee0458b1452f8be4502807ff78109ae4b0fa821

Comment 8

[Andrew McCreight [:mccr8]]

Yeah, I just finished rerunning a mozregression and I got the same thing as comment 7. :)

Comment 11

[:Felipe Gomes (needinfo for replies!)]

(In reply to Andrew McCreight [:mccr8] from comment #8)
> Yeah, I just finished rerunning a mozregression and I got the same thing as
> comment 7. :)

Me too. It was hard to bisect because it failed about 70% of the time only.. so at times it threw the bisection in the wrong direction.. After some careful retries I arrived here too.

One easy website to test this is imgur.com. It supports drag'n'dropping of image files to the homepage

Comment 13

[Ryan VanderMeulen [:RyanVM]]

Today's nightlies contain the backout of bug 1351148, so this should be fixed now.

Comment 14

[Timea Zsoldos [:zstimi/tzsoldos], Desktop QA]

I have reproduced this issue using Firefox 56.0a1 (2017.07.24) on Win 8.1 x64.
I can confirm this issue is fixed, I verified using Firefox 56.0b5 Win 8.1 x64, Mac OS X 10.12 and Ubuntu 14.04 x64.