> .hgignore:
>
> This seems to be bug 4077 which Comay has just putback recently.
Ok. That should disappear in the next merge.
> filelist.py:
> line 509: s/. Decompress/. Decompress/
Ok.
> lines 528, 542: suggestion: s/should be/expected/
Ok.
> misc.py:
> line 517: a docstring would be nice
None of the other exceptions had a docstring. I added one to each
TransportException so this one doesn't look so oddly lonely.
> line 551: two periods ("%s. %s.")? really?
Yes. Take a look at filelist:_verify_content(). The exception data is
a string and none of them have periods. We actually do something
weirdly similar in TransferContentException and
TransferTruncatedException. That said, this should be modified so it
doesn't print %s if e.data is none.
> retrieve.py:
> line 127-149: I'm not sure I'm keen on code being aware of updatelog
> being in retrieve. My inclination is to only have network retrieval
> code in retrieve.py; but I'm not deadset on this.
What alternative are you proposing? The retrieve code is
client-specific, so we're not breaking any boundaries between the client
and server. Given that the exception handling code between filelist,
catalog, manifest, and datastream are all very similar, it seemed
sensible to start centralizing the routines there, at least for now.
This is all going to change once we re-write the transport stuff.
> updatelog.py:
> line 272: given line 276 is this still right?
276 is just skipping lines where we don't recognize the format. The
content-length header should report the size of the entire message. If
we don't include the size of skipped lines in the total, there's a
slight possibility of a mismatch at the end.
Thanks for reviewing this.
-j
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
