Between Mercurial 1.8.1 and 1.8.2 this change was included: changeset: 13693:adf3c4401c5d branch: stable user: Miloš Hadžić <milos.hadzic@gmail.com> date: Thu Mar 17 22:55:02 2011 +0100 summary: push/outgoing: print remote target path even if there's an error (issue2561) With python 2.5 and 2.4 (not with 2.6) it produces a failed test: --- /home/thomas/hg/repos/tah/tests/test-ssh.t +++ /home/thomas/hg/repos/tah/tests/test-ssh.t.err @@ -266,7 +266,7 @@ passwords in ssh urls are not supported $ hg push ssh://user:erroneouspwd@dummy/remote - pushing to ssh://user:***@dummy/remote + pushing to ssh://user:erroneouspwd@dummy/remote abort: password in URL not supported! [255] The failed test only occurs in the stable branch, not with default (tested with 1aea86673dee) Since the problem only affects ssh:// (replacing the password with *** works with http and https, even if I provide an invalid host name) and passwords in URLs are not supported for ssh, I do not consider this critical.
Is this really urgent and worth fixing on the stable branch? Hiding of ssh passwords never worked - Miloš just made it a bit more visible. It is an error in the first place to specify the password this way, so this error message should only be hit the one and only time a user tries to do this. Anyway: This could be another reason for backporting Brodies url module and using it in a few places. (As you said on Issue2556 ...)
The urgent part is that a test is broken. While a broken test is no big problem in itself, it leads to ignoring broken tests which leads to missing important problems. Adjusted title to reflect this.
Agreed. Would it be fine with you to just add a (glob) in stable?
* Mads Kiilerich <bugs@mercurial.selenic.com> [20110406 15:13]: > Would it be fine with you to just add a (glob) in stable? fine with me.
I've got a fix for this in my tree already.
--- Bug imported by bugzilla@serpentine.com 2012-05-12 09:19 EDT --- This bug was previously known as _bug_ 2754 at http://mercurial.selenic.com/bts/issue2754