569509 - TinderboxPushlog Robot has stopped working because of UTF-8 nbsp char in the json which can't be handled by the server

Status: RESOLVED FIXED

(Tree Management Graveyard :: TBPL, defect)

Description

[(no longer active)]

Attachment: Patch (v1)

The simplest workaround here is to replace UTF-8 nbsp chars in the input stream with normal spaces before JSON encoding, as this patch does.

Markus, it would be great if you can land and deploy this patch, so that tbplbot returns asap.  Thanks!

Comment 1

[Markus Stange [:mstange]]

I haven't understood this patch yet, but I pushed and deployed anyway,
http://hg.mozilla.org/users/mstange_themasta.com/tinderboxpushlog/rev/8072a93dd06f

Comment 2

[(no longer active)]

Thanks!

The UTF-8 nbsp sequence (0xc2a0) is converted into \u00a0 in the json string, and apparently the json parser that the bugzilla API uses can't handle that, so we'll get an error from the bugzilla API server.  This didn't use to be an issue when we used GET because that character was escaped to 0xa0 when being passed as the URI.

Comment 3

[Markus Stange [:mstange]]

Where do these nbsp chars enter our world? Where do they come from?

Comment 4

[(no longer active)]

(In reply to comment #3)
> Where do these nbsp chars enter our world? Where do they come from?

The Tinderbox log summary, between the date and the time.  (At least that's one place they exist, not sure if those are the only one.)

Comment 5

[Markus Stange [:mstange]]

Comment on attachment 448664 [details] [diff] [review]
Patch (v1)

Ah ok, cool, thanks.