Changes between Initial Version and Version 1 of Ticket #1534, comment 5


Ignore:
Timestamp:
2011-09-18T17:47:31Z (13 years ago)
Author:
davidsarah
Comment:

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #1534, comment 5

    initial v1  
    22> The caller in publish handles the case in which the remote write fails (e.g., because of an uncoordinated write error, share corruption, or something else). It's possible that we don't have tests to exercise that, or that the exception is getting lost in the deferred chain somewhere. I'll take a look at it.
    33
    4 It seems that both of these contribute to the missing exception. attachment:1534-dropped-error-and-tests.darcs.patch (or attachment:1354-dropped-error-and-tests.darcs.patch, if you can tolerate the typo :-) tweaks the control flow in the publisher a little bit so the exceptions don't get hidden, and then adds a test to exercise the code for MDMF files. {{{UnknownVersionError}}} still doesn't get raised, though, since the MDMF checkstring is shorter than the SDMF checkstring and causes {{{struct.unpack}}} to give up before the version can be extracted and checked. I'll work on a patch to make the new test pass.
     4It seems that both of these contribute to the missing exception. attachment:1534-dropped-error-and-tests.darcs.patch tweaks the control flow in the publisher a little bit so the exceptions don't get hidden, and then adds a test to exercise the code for MDMF files. {{{UnknownVersionError}}} still doesn't get raised, though, since the MDMF checkstring is shorter than the SDMF checkstring and causes {{{struct.unpack}}} to give up before the version can be extracted and checked. I'll work on a patch to make the new test pass.
     5
     6[edit: remove reference to duplicate patch]