1033313 - Can't import some patches

Status: NEW

(Developer Services :: Mercurial: qimportbz, defect)

Description

[:Gijs (he/him)]

STR:

gkruitbosch-16516:fx-team gkruitbosch$ hg qimport bz://1018154
Fetching... done
Parsing...Patch id=8448680 desc="Part 2, v1: test changes" diff data were discarded:
UnicodeDecodeError: 'utf8' codec can't decode byte 0xe9 in position 5191: invalid continuation byte
Patch id=8448712 desc="Part 2, v1.1: test changes" diff data were discarded:
UnicodeDecodeError: 'utf8' codec can't decode byte 0xe9 in position 5446: invalid continuation byte
 done
1: Part 2, v1.1: test changes
  bmcbride: review+
2: Part 3, v1: new test for the paste functionality
  bmcbride: review-
3: Part 1, v1.2: code changes
  dao: review+
Which patches do you want to import? [Default is '1', use '1-3' for all] 1
adding 1018154 to series file
renamed 1018154 -> 1018154-tests
gkruitbosch-16516:fx-team gkruitbosch$ hg qpush
applying 1018154-tests
patch 1018154-tests is empty
now at: 1018154-tests

:-(

(to be clear, I'd have expected no errors and the patch shouldn't be empty)

Comment 1

[Ed Morley [:emorley]]

Fwiw I get the same for that bug:

[/c/src-gecko/inbound]$ qimp 1018154
Fetching... done
Parsing...Patch id=8448680 desc="Part 2, v1: test changes" diff data were discarded:
UnicodeDecodeError: 'utf8' codec can't decode byte 0xe9 in position 5191: invalid continuation byte
Patch id=8448712 desc="Part 2, v1.1: test changes" diff data were discarded:
UnicodeDecodeError: 'utf8' codec can't decode byte 0xe9 in position 5446: invalid continuation byte
 done
...

Whereas for other bugs it works fine, eg:

[/c/src-gecko/inbound]$ qimp 1027004
Fetching... done
Parsing... done
adding 1027004 to series file


[/c/src-gecko/inbound]$ hg --version
Mercurial Distributed SCM (version 3.0.1)

[/c/src/vctools]$ hg parent
changeset:   489:b857cebfcddc
tag:         tip
user:        Gregory Szorc <gps@<snip>.com>
date:        Tue Jul 01 20:27:08 2014 -0700
summary:     reviewboard: refactor unused code

Comment 2

[(no longer active)]

In the git bz extension we had a similar bug caused by us trying to encode the attachment into UTF-8.  The fix is to not change the encoding of the attachment when downloading it.