WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
Bug 194966
Web Inspector: Dark Mode: Network Overview Graph segments have distracting white box shadow
https://bugs.webkit.org/show_bug.cgi?id=194966
Summary
Web Inspector: Dark Mode: Network Overview Graph segments have distracting wh...
Joseph Pecoraro
Reported
2019-02-22 16:19:14 PST
Dark Mode: Network Overview Graph segments have distracting white box shadow The box shadow is supposed to be the background of the timeline (white in light mode) and needs to be updated for dark mode.
Attachments
[PATCH] Proposed Fix
(1.85 KB, patch)
2019-02-22 16:20 PST
,
Joseph Pecoraro
no flags
Details
Formatted Diff
Diff
[IMAGE] Before
(321.46 KB, image/png)
2019-02-22 16:20 PST
,
Joseph Pecoraro
no flags
Details
[IMAGE] After
(321.00 KB, image/png)
2019-02-22 16:20 PST
,
Joseph Pecoraro
no flags
Details
[PATCH] Proposed Fix
(6.69 KB, patch)
2019-02-22 16:48 PST
,
Joseph Pecoraro
no flags
Details
Formatted Diff
Diff
[PATCH] Proposed Fix
(6.68 KB, patch)
2019-02-22 16:51 PST
,
Joseph Pecoraro
no flags
Details
Formatted Diff
Diff
[PATCH] Proposed Fix
(5.92 KB, patch)
2019-02-22 18:45 PST
,
Joseph Pecoraro
hi
: review+
Details
Formatted Diff
Diff
Show Obsolete
(3)
View All
Add attachment
proposed patch, testcase, etc.
Joseph Pecoraro
Comment 1
2019-02-22 16:20:05 PST
Created
attachment 362779
[details]
[PATCH] Proposed Fix
Joseph Pecoraro
Comment 2
2019-02-22 16:20:20 PST
Created
attachment 362780
[details]
[IMAGE] Before
Joseph Pecoraro
Comment 3
2019-02-22 16:20:34 PST
Created
attachment 362782
[details]
[IMAGE] After
Nikita Vasilyev
Comment 4
2019-02-22 16:31:55 PST
Comment on
attachment 362779
[details]
[PATCH] Proposed Fix View in context:
https://bugs.webkit.org/attachment.cgi?id=362779&action=review
> Source/WebInspectorUI/UserInterface/Views/NetworkTimelineOverviewGraph.css:49 > + .timeline-overview-graph.network > .graph-row > .timeline-record-bar > .segment:not(.inactive) { > + box-shadow: hsl(0, 0%, 21%) 1px 0 0;
This is fine, but we could also use a variable: .timeline-overview-graph.network > .graph-row > .timeline-record-bar > .segment:not(.inactive) { --box-shadow-color: white; box-shadow: var(--box-shadow-color) 1px 0 0; } @media (prefers-color-scheme: dark) { .timeline-overview-graph.network > .graph-row > .timeline-record-bar > .segment:not(.inactive) { --box-shadow-color: hsl(0, 0%, 21%); } } Same for the 2nd rule.
Joseph Pecoraro
Comment 5
2019-02-22 16:48:49 PST
Created
attachment 362787
[details]
[PATCH] Proposed Fix Use variables now.
Joseph Pecoraro
Comment 6
2019-02-22 16:50:18 PST
Comment on
attachment 362787
[details]
[PATCH] Proposed Fix View in context:
https://bugs.webkit.org/attachment.cgi?id=362787&action=review
> Source/WebInspectorUI/UserInterface/Views/CPUTimelineOverviewGraph.css:57 > + background-color: var(--timeline-alternate-background-color);
Typo this should be --timeline-even.
Joseph Pecoraro
Comment 7
2019-02-22 16:51:23 PST
Created
attachment 362788
[details]
[PATCH] Proposed Fix
Nikita Vasilyev
Comment 8
2019-02-22 16:54:41 PST
Patch doesn't apply to ToT.
Joseph Pecoraro
Comment 9
2019-02-22 18:45:40 PST
Created
attachment 362811
[details]
[PATCH] Proposed Fix
Devin Rousso
Comment 10
2019-02-22 20:14:47 PST
Comment on
attachment 362811
[details]
[PATCH] Proposed Fix View in context:
https://bugs.webkit.org/attachment.cgi?id=362811&action=review
> Source/WebInspectorUI/UserInterface/Views/Variables.css:117 > + --timeline-odd-background-color: var(--background-color-content);
I don't think we want to do this for light-mode. We should keep it as `white`. making it match the background color will cause it to "blend in" a bit. I found that a border/shadow with this color almost made it seem like records were getting cut in half by other records (e.g. in the overview), rather than being laid on top of each other.
> Source/WebInspectorUI/UserInterface/Views/Variables.css:118 > + --timeline-even-background-color: hsl(0, 0%, 96%);
Even and odd are weird choice of name. How about "primary" and "alternate"? I get that in these cases it makes sense (e.g. `:nth-child(even)` or `:nth-child(odd)` are actually used), but to someone just looking at Variables.css it might seem confusing. At the very least, more description as to what is even/odd (e.g. legend items, or network record "bubbles") would be great.
Matt Baker
Comment 11
2019-02-25 09:57:33 PST
(In reply to Devin Rousso from
comment #10
)
> Comment on
attachment 362811
[details]
> [PATCH] Proposed Fix > > View in context: >
https://bugs.webkit.org/attachment.cgi?id=362811&action=review
> > > Source/WebInspectorUI/UserInterface/Views/Variables.css:117 > > + --timeline-odd-background-color: var(--background-color-content); > > I don't think we want to do this for light-mode. We should keep it as > `white`. making it match the background color will cause it to "blend in" a > bit. I found that a border/shadow with this color almost made it seem like > records were getting cut in half by other records (e.g. in the overview), > rather than being laid on top of each other. > > > Source/WebInspectorUI/UserInterface/Views/Variables.css:118 > > + --timeline-even-background-color: hsl(0, 0%, 96%); > > Even and odd are weird choice of name. How about "primary" and "alternate"? > I get that in these cases it makes sense (e.g. `:nth-child(even)` or > `:nth-child(odd)` are actually used), but to someone just looking at > Variables.css it might seem confusing. At the very least, more description > as to what is even/odd (e.g. legend items, or network record "bubbles") > would be great.
The words even and odd also match the system semantic color names: -apple-system-even-alternating-content-background -apple-system-odd-alternating-content-background
Joseph Pecoraro
Comment 12
2019-02-25 10:55:55 PST
(In reply to Devin Rousso from
comment #10
)
> Comment on
attachment 362811
[details]
> [PATCH] Proposed Fix > > View in context: >
https://bugs.webkit.org/attachment.cgi?id=362811&action=review
> > > Source/WebInspectorUI/UserInterface/Views/Variables.css:117 > > + --timeline-odd-background-color: var(--background-color-content);
>
> I don't think we want to do this for light-mode. We should keep it as > `white`. making it match the background color will cause it to "blend in" a > bit.
This doesn't change things. --background-color-content is already white in light mode. UserInterface/Views/Variables.css 50: --background-color-content: white;
> > Source/WebInspectorUI/UserInterface/Views/Variables.css:118 > > + --timeline-even-background-color: hsl(0, 0%, 96%); > > Even and odd are weird choice of name. How about "primary" and "alternate"? > I get that in these cases it makes sense (e.g. `:nth-child(even)` or > `:nth-child(odd)` are actually used), but to someone just looking at > Variables.css it might seem confusing. At the very least, more description > as to what is even/odd (e.g. legend items, or network record "bubbles") > would be great.
I wasn't aware that this matches the -apple-system naming pattern, but I also support the odd/even. primary/alternate is fine but odd/even makes it very obvious that during a `*:nth-child(even)` that you're going to use the even variant and not look up what to use.
Devin Rousso
Comment 13
2019-02-25 13:45:44 PST
Comment on
attachment 362811
[details]
[PATCH] Proposed Fix View in context:
https://bugs.webkit.org/attachment.cgi?id=362811&action=review
rs=me
> Source/WebInspectorUI/UserInterface/Views/NetworkTimelineOverviewGraph.css:40 > + box-shadow: var(--timeline-odd-background-color) 1px 0 0;
Rather than repeat the `box-shadow` declaration, you could have another variable that just defines the color, and just change that variable when `:nth-child(even)`. box-shadow: var(--active-segment-box-shadow-color) 1px 0 0; --active-segment-box-shadow-color: var(--timeline-odd-background-color); --active-segment-box-shadow-color: var(--timeline-even-background-color);
> Source/WebInspectorUI/UserInterface/Views/Variables.css:285 > + --timeline-even-background-color: hsl(0, 0%, 25%);
I think we should add a comment here explaining that if `--background-color-content` gets changed, this should also be adjusted to match.
Joseph Pecoraro
Comment 14
2019-02-25 20:10:26 PST
> I think we should add a comment here explaining that if > `--background-color-content` gets changed, this should also be adjusted to > match.
If the odd color changes then we should probably adjust the even color, yes. I don't think a comment will help with that much. More likely if we (very unlikely) change --background-color-content we should see that we are changing the odd color. even may not need to change at all.
Joseph Pecoraro
Comment 15
2019-02-25 22:37:41 PST
https://trac.webkit.org/r242072
Radar WebKit Bug Importer
Comment 16
2019-02-25 22:40:49 PST
<
rdar://problem/48391381
>
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug