In "What happens instead?", I meant to say that "The field values are *supposed* to be included as POST params when submitting the form."

Will it be possible for you to provide a sample embedded form for testing purpose?

Sure! I've copied the HTML of a sample embedded form and placed it in this jsfiddle: https://jsfiddle.net/rhopd162/

After filling out the form and hitting submit, a new tab should open saying that your response has been recorded. You should also see a link that says "See previous responses". Clicking on this link shows all of the response values submitted for this form thus far.

If you submit responses to the embedded form in Chrome 54.0.2840.41 beta or 55.0.2873.0 dev, you'll see that the number of responses goes up on the "See previous responses" page, but no new response values will appear for either question. However, if you submit a response using Chrome 53.0.2785.143, and then click on "See previous responses", you'll see that the response values you just entered will be present there.

Let me know if you run into any issues with the sample!

Able to reproduce the issue on Ubuntu 14.04 using reported versions #54.0.2840.41 (beta) and #55.0.2873.0 (dev) and latest canary #55.0.2878.0

Bisect Information:
=====================
Good build: 54.0.2807.0
Bad Build : 54.0.2808.0

Change Log URL: 

https://chromium.googlesource.com/chromium/src/+log/98fd60a637e0cd2609cb8fb909503bb6cb9b40c6..3e3be13746968b4e07c0ffdc707b05a14676e435

From the above change log suspecting below change

Review url: https://codereview.chromium.org/2165133003


lukasza@ - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

kevinx@, this is possibly a duplicate of  issue 646261 .  I'll try to repro this myself later today, but to make sure that I did not misunderstand the repro steps, it would be great if you could check if you can repro in the latest canary (55.0.2875.0 or later;  the fix has been merged to M54/beta but AFAIK no beta build with the fix has been released yet).

Hey lukasza@, I am happy to report that I cannot repro this issue in canary build 55.0.2879.0 on OSX.

Is it possible for me to get the canary build on Ubuntu as well for testing? https://www.google.com/chrome/browser/canary.html says Chrome canary is unavailable on linux platforms. 

There is no official release channel for Chrome canaries on Linux (only on Windows and Mac), but in addition to daily canaries, the trunk also designates one of the canaries as a dev channel build (this happens around once a week).  The dev channel is available on Linux as google-chrome-unstable package.  Unfortunately, dev channel is currently on 55.0.2873.0 (so doesn't yet include the fix for  issue 646261 ).

No problem, was just curious. I'll test on linux once the dev channel goes past 55.0.2875.0.

Hmmm... I tried to repro before and after the CL with the fix in M55 (i.e. at 1d02573^ and 1d02573) and unfortunately I couldn't repro in both cases.  My first guess would be that the repro depends on some extra extensions...

Random thoughts / notes:

- #c8 above (blaming r407586) is consistent with the hypothesis that this is a duplicate of  issue 646261 

- I don't see any custom enctype / Content-Type in the form - this seems unlikely to be  issue 648648 

- Given that the bug doesn't repro anymore on the latest canary (as reported in #c12) I think we can tentatively close as a duplicate.  We should try to repro again once Beta and/or Dev channels are refreshed to include the fix.

- We will probably need to make some extra tweaks around ShouldFork / POST in M56 - I've added a comment to issue 650694 to be mindful of the Google Forms scenario raised here.

For what it's worth, I'm able to repro the issue on the current Chrome dev build (55.0.2873.4) on OSX. However, the issue does not repro for me on the canary build (55.0.2879.0) on OSX.

Hey lukasza@, do you know when the Beta and/or Dev channels for Linux will be refreshed to include the fix? I checked https://www.chromium.org/getting-involved/dev-channel but it doesn't seem like any new releases have gone out yet.

The fix (r421645) should be present in 55.0.2875.0 and later + also was merged into 54.0.2840.45 and later.  So, AFAICT, currently both the dev and beta channels should have the fix (I think they are currently on 55.0.2882.4 and 54.0.2840.50).

Ah, my mistake - I didn't realize the fix had been merged into 54.0.2840.45 and later. I can confirm that the bug no longer repros on either the dev or beta channels for Linux. Thanks for all your help lukasz@!