Loggerhead URL handling brittle
John Arbash Meinel
john at arbash-meinel.com
Tue Jan 13 15:30:55 GMT 2009
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Matt Nordhoff wrote:
> Robert Collins wrote:
>> On Tue, 2009-01-13 at 04:02 +0000, Matt Nordhoff wrote:
>>> "" seems to be a valid revision ID, so Loggerhead's behavior is
>>> correct,
>>> if not very useful.
>> It's not, if bzr isn't complaining its a bug.
>>
>> -Rob
>
> Oh. Well, I don't know if it complains. I can't get the command line to
> complain about anything, and I don't know bzrlib. I asked in #bzr, and
> the answer was a tentative "it's probably allowed".
So... the *read* functions generally don't do any assertion about the
revision strings. The *write* functions are designed to forbid certain
constructs. I'm not sure about the empty string, but they certainly
forbid things like whitespace in the string, and a trailing ':' is
considered to be an internal revision marker that shouldn't be committed
to disk (current:, null:, etc)
I would expect that you can't get "bzr log -r revid:XXXX" to fail for
any combination of XXXX because it assumes that if it is invalid, it
just won't match because we wouldn't allow you to commit an invalid XXXX.
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEARECAAYFAklss68ACgkQJdeBCYSNAAO30ACdH6dXpeyQxTiof4lUVZi2vSva
0/UAoL1/ByvK1NheNkgbW5m+glLCjXZF
=7Oun
-----END PGP SIGNATURE-----
More information about the bazaar
mailing list