Closed GoogleCodeExporter closed 9 years ago
Patch committed at R138 (part of the patch is also customization of the
"phaseInterval" through a cookie), doc and test needed yet.
Thanks for the report!
Will be in 1.1-11.
Honza
Original comment by odva...@gmail.com
on 16 Aug 2010 at 6:15
Another schema update (also related to pageTimings.onContentLoad) committed at
R139
Honza
Original comment by odva...@gmail.com
on 24 Aug 2010 at 3:03
Selenium test for this issue committed at R141
Honza
Original comment by odva...@gmail.com
on 24 Aug 2010 at 3:21
HAR Viwer 1.1-11 released
Honza
Original comment by odva...@gmail.com
on 24 Aug 2010 at 3:33
Doc updated
http://code.google.com/p/harviewer/wiki/Customization
Honza
Original comment by odva...@gmail.com
on 24 Aug 2010 at 3:51
Thank you very much
HARViewer now accept har without onLoad field, great :)
But the second timings line start at 0ms, and should start after the first line
(this is the behaviour with the onLoad timing).
See the difference between
- withMinusOneLoad.har
- withOnLoad.har
Thank you
Original comment by filir...@gmail.com
on 13 Sep 2010 at 12:54
Attachments:
> But the second timings line start at 0ms, and should start after the
> first line (this is the behaviour with the onLoad timing).
I don't understand what you mean.
> See the difference between
> - withMinusOneLoad.har
> - withOnLoad.har
withOnLoad.har - seems to be fine for me. I see both vertical lines
(onContentLoad, onLoad)
withMinusOneLoad.har - this seems to be wrong since onContentLoad is in HAR and
not displayed. Is this the problem, you have in mind?
Honza
Original comment by odva...@gmail.com
on 13 Sep 2010 at 4:57
No, I was not clear.
I joinded a screenshot to make it clearer.
with onLoad, the timing line looks like this :
html : ---
css : --------
witout onLoad, the timing line looks like this :
html : -----------
css : -----------
And we all know that the css start after the html.
Original comment by filir...@gmail.com
on 14 Sep 2010 at 7:45
Attachments:
Ah, understand now. This is actually not a bug, but rather a feature.
If a time lag between two HTTP requests (counted from the end of the first one
till the beginning of the second one) exceeds a limit, the graphical timeline
breaks layout and requests are drawn again from the beginning (left side) of
the waterfall diagram.
This allows better utilization of the time axes (horizontal) (in case of large
lags).
Some more info here:
http://www.softwareishard.com/blog/firebug/firebug-net-panel-timings/
Anyway, there is an exception, all requests executed during the page load phase
(till onload event) are always displayed together (there is no break even if
the gap exceeds the limit).
In your case, the gap between the two requests is bigger than 1000ms (the
default value) and the reason why they are displayed together is the "onload"
event which represents the end of page load - if it's missing then the
layout-break happens.
There are two solutions for you:
1) make the default limit bigger. Create phaseInterval cookie and set it to
e.g. 5000
(you can use Firecookie - http://www.softwareishard.com/blog/firecookie/)
See other cookies:
http://code.google.com/p/harviewer/wiki/Customization
2) Latest (but still debug) version of HAR Viewer allows to customize these
"layout
See a context menu at the end of the request row and pick "Break Timeline
Layout" (it's checked if the break happened at the clicked request).
See e.g. this HAR example:
http://www.janodvarko.cz/har/viewer-debug/?path=examples/www-cnn-com.har
click at the 6th request and remove the break layout mark.
This is new, feedback welcome!
Honza
Original comment by odva...@gmail.com
on 14 Sep 2010 at 8:18
Perfect, this is exactly what I need.
Maybe later, I will try to disable the break time line by deafult.
Thanks
Romain
Original comment by filir...@gmail.com
on 14 Sep 2010 at 9:05
Original issue reported on code.google.com by
filir...@gmail.com
on 9 Aug 2010 at 3:30Attachments: