We should avoid this spurious output.
Created attachment 226152 [details]
This issue was fixed in <http://trac.webkit.org/changeset/165541> (bug #130129). Maybe it would be better to use which(1) to detect whether ninja is in the user's path as opposed to calling ninja and redirecting stderr (thereby silencing the "command not found" error) as in the aforementioned change set?
Comment on attachment 226152 [details]
It is really better fix than r165541, r=me.
(but cq-, because it is in conflict with r165541.)
I think this is a good bug to switch to 'which' and also to detect ninja-build as well (as it is called on Fedora).
Created attachment 226790 [details]
Comment on attachment 226790 [details]
Clearing flags on attachment: 226790
Committed r165684: <http://trac.webkit.org/changeset/165684>
All reviewed patches have been landed. Closing bug.