HAR export - incorrect startedDateTime

Support
1

I noticed that “startedDateTime” values in the JSON HAR export are missing timestamp values which are required to generate nice waterfall views in HAR reader tools. Without this data the waterfalls are visually flat.

startedDateTime":“2012-03-16T07:39:28[color=#FF0000].000+00:00[/color]”
startedDateTime":“2012-03-16T07:39:38[color=#FF0000].000+00:00[/color]”

The applicable test result can be found here:
http://www.webpagetest.org/result/120316_ZQ_3KRWS/

Is this a bug? Or is the data not included on purpose?

Thanks in advance.

2

It is a bug. I thought I had an issue open on it but someone must have mailed it to me. It’s on my list of things to get fixed really soon (within the next week) : Google Code Archive - Long-term storage for Google Code Project Hosting.

3

Great to have it fixed soon, thanks!

4

ok, the times have been fixed - sorry about that (we had changed a variable name and missed updating it in one place). The online HAR viewer seems to be jumbling up the order of the requests (not sure why but I looked at the raw JSON and they are fine).

Let me know if you see anything else odd going on there.

Thanks,

-Pat

5

Just checked the HAR output generated via this url: http://www.webpagetest.org/export.php?test=120326_3R_3QH5T and the output still misses the relevant timestamp values.

Can you have a look please?
Thanks

6

Any chance you are getting a cached version of the file from your browser? I just pulled it and loaded the result into the HAR viewer and got a waterfall. Looking at the first 4-5 requests in the raw JSON file look like they are correctly tracking the startedDateTime:

“startedDateTime”:“2012-03-26T01:45:54.482+00:00”
“startedDateTime”:“2012-03-26T01:45:55.334+00:00”
“startedDateTime”:“2012-03-26T01:45:55.334+00:00”
“startedDateTime”:“2012-03-26T01:45:55.800+00:00”
“startedDateTime”:“2012-03-26T01:45:55.920+00:00”

Thanks,

-Pat

7

You are right, my HAR reader was fooling me :wink: Har viewer is showing the right waterfall.

Thanks for the double check!

8

Would this fix be available in hosted version? 2.6 has exactly the same issue.

9

Yes, it will be in the next release or you can grab the fix from svn: http://webpagetest.googlecode.com/svn/trunk/www/webpagetest/object_detail.inc

if you want to be even more surgical, you can just change ‘offset’ to ‘load_start’ in LoadRequests()