BarnOwl-to-other-AIM-client encoding problem

[zhangc@mit.edu]

There seems to be a bug with the BarnOwl-AIM interface, although I haven't really figured out whether this BarnOwl's or AIM's fault.

I sent a set of test messages (img1) from one client to another (set of clients: BarnOwl v1.3, GMail AIM client, Pidgin v2.4.2).

Sending messages from BarnOwl to the GMail AIM client:
Test message #0 (TM0) returns the input
TM1 returns nothing
TM2 returns the input
TM3 returns nothing
TM4 returns "<"
TM5 returns an empty line
TM6 returns line 1 of img2
TM7 returns lines 2 and 3 of img2
TM8 returns line 4 of img2
TM9 returns line 5 of img2

Sending messages from BarnOwl to Pidgin:
TM0 returns the input
TM1 returns an empty line
TM2 returns the input
TM3 returns the input
TM4 returns "<"
TM5 returns line 1 of img3
TM6 returns line 2 of img3
TM7 returns line 3 of img3
TM8 returns line 4 of img3
TM9 returns line 5 of img3

Sending messages from the GMail AIM client *OR* Pidgin to BarnOwl:
TM0 returns the input
TM1 returns the input
TM2 returns the input
TM3 returns the input
TM4 returns the input
TM5 returns the input
TM6 returns "&#4e2d;&#6587;"
TM7 returns "&#65e5;&#672c;&#8a9e;&#3068;&#3072;&#3089;&#304c;&#306a;"
TM8 returns "&#c0ac;&#b791;"
TM9 returns the input
(These are not actually escaped HTML entities, although s/#/#x/ results in readable HTML entities.)

Sending/receiving between Pidgin and the GMail AIM client always returns the input.

[zhangc@mit.edu]

My images don't seem to want to be uploaded, so I've put them in my Public for the time being: web/zhangc/Public/barnowl/

[nelhage@mit.edu]

I believe I have this fixed on the 'unicode-fixes' branch (sending and receiving to both pidgin and gmail now work for me), and this should be fixed in the next release.

[nelhage@mit.edu]

Hm. I realized I haven't tested the HTML meta-character ones, only the unicode ones. Presumably we also need to be doing HTML escaping on outgoing messages, as well, I'll get to that and test it sometime as well.

[nelhage@mit.edu]

The unicode issue was fixed elsewhere. I'll open a new ticket about the HTML escaping.