Identify automatic str/unicode coercions

[Martin von Gagern]

John Arbash Meinel wrote:
> I think using "warnings.warn" is a reasonable place to do this. Come up 
> with
> your own warning class, and it can be easily filtered. And mostly, since we
> shouldn't be doing automatic conversions, this log should dwindle with 
> time.
> 
> John
> =:->

While this seems appropriate in the long run, for the time being I've 
instead implemented my SqLite-based idea. Catch up to the latest 
development on https://code.launchpad.net/~gagern/bzr/str-unicode

I found that figuring out whether the string to be converted originated 
in a literal on the stack is quite a tricky thing to do, especially for 
whitespace and similar non-alphanumeric strings. So those are still part 
of the log, and while they pose no danger for bzr behaviour, converting 
these literals to the correct type would be cleaner nevertheless.

Is there anybody willing to grab one of these reports and tackle the 
conversions one by one?

Greetings,
  Martin von Gagern


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 260 bytes
Desc: OpenPGP digital signature
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20080617/1795d06a/attachment.pgp