cvs source UnicodeDecodeError

[anonymous]

Problem description

I've been trying to import multiple cvs projects using tailor rev [1075]. Some works, but some fails with the following error while bootstrapping HEAD.

• symptoms

    File "/usr/lib/python2.4/codecs.py", line 290, in read
      newchars, decodedbytes = self.decode(data, self.errors)
  UnicodeDecodeError: 'ascii' codec can't decode byte 0xf1 in position 28: ordinal not in range(128)
  

... hum obviously the problem here is the ascii decoding function is too strict with cvs repositories where the rlog contains some invalid chars.

---

Quick fix

vcpx/repository.py @ lines 100,101

• Original Source

          self.encoding_errors_policy = cget(self.name,
                                             'encoding-errors-policy', 'strict')
  

• Modified source

          self.encoding_errors_policy = cget(self.name,
                                             'encoding-errors-policy', 'ignore')
  

solved the problem for me, but i guess this might apply only as a temporary solution... (sorry i'm no python expert).

---

btw ... thx for this great tool !

[Lele Gaifax]

Isn't it enough to set the ignore value in the configuration?

[anonymous]

Hi lele,

i guess so,

sorry i did not found out earlier, so, for people ending up on this page the following line can be added to your configuration file :

encoding-errors-policy = ignore

Could the code, either point out the error, and/or output the correct configuration sequence, and/or catch it by hitself and adapt seamlessly ?

thx.

[Lele Gaifax]

Yes, that would be nicer. It should probably be enough to catch also the UnicodeDecodeError exception in vcpx/tailor.py@1065#L145.

[lele]

Fixed by [1080].