1366091 - Voltage SecureMail (HPE SecureMail) is broken in nightly when email message loaded as file: URL

Status: VERIFIED DUPLICATE of bug 1351358

(Core :: General, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Voltage SecureMail, which is a product described at https://www.voltage.com/products/email-security/hpe-securemail/ , is broken in nightly but not in beta.  However, based on bisecting, it's been broken in nightly for quite a while, since this one day range:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=bad312aefb42982f492ad2cf36f4c6c3d698f4f7&tochange=f8f4eaac1701107f794b48891bcca2c95d39d503

My steps to reproduce are:
 * load the HTML email (launched from my email client) in Firefox as a file: URL
 * click the [Click to Read Message] button

Actual results:

Page showing:
> [company name] SecureMail Error
> There was a problem servicing your request. Please try again later.
> 
> Click here for additional information.
> 
> Error Code: 23

Expected results:

Page showing:

> [SECURE DOCS]
> 
> Create a password to continue
> 
>    Full Name
> [etc...]

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Toggling the preference "browser.tabs.remote.separateFileUriProcess" to false fixes it.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Regression from the pref added in https://hg.mozilla.org/mozilla-central/rev/fe6ea50bfb38e8e9a93408a5f9b3c26082073ecb

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

The thing that this the HTML that's loaded at a file: URL is doing is that it has a <form method="post"> to post to an https: URL, with about 2 megabytes of data in a series of 761 <input type="hidden"> elements (plus a few other <input type="hidden"> with metadata, plus the single <input type="button"> to submit the form.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

(I suspect this may have something to do with the size of the POST data involved being sent over IPC...)

Comment 5

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Actually, I think this has nothing to do with the size -- it seems like we're turning POSTs into GETs incorrectly.  I'll attach a testcase using a public site.

Comment 6

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: simple testcase (needs to be loaded from file: URL to show failure)

This posts to the post endpoint of https://httpbin.org/ .  If it works correctly, you'll see a visual representation of the JSON results.  If it fails, you'll see:

> Method Not Allowed
> 
> The method is not allowed for the requested URL.

Comment 8

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

And I've confirmed that this is fixed when I update my tree to mozilla-inbound from a few hours ago (with the changes from bug 1351358).