Hi Mark,

Your changes look good to me.

Thanks,
-Suha

On 06/15/09 15:22, Mark J. Nelson wrote:
The impetus for this change is described in Alan's note:
http://mail.opensolaris.org/pipermail/tools-discuss/2009-April/004513.html

The two webrevs:

http://cr.opensolaris.org/~mjnelson/webrev.arcdb/
http://cr.opensolaris.org/~mjnelson/webrev.arcdb_tests/

Outside of DbLookups.py, the remaining changes (and test changes) all
involve getting rid of the no-longer-valid assumption that arc case
titles are truncated to 40 characters.

Stuff I wouldn't mind a sanity check on:

- Do the tests need to explicitly force off-SWAN lookups?  I chose "no,"
allowing them to implicitly use whichever data source they could access.
 This is different from the Monaco/BOO case, I think, because the
underlying interfaces are the same.  Whereas, for Monaco vs BOO, the
DbLookups code is significantly different.

- Treating "error" and "absent" identically: "error" generally indicates
a badly formed request, and "absent" indicates a well formed request for
an ARC case that does not exist.  Since the return is simply a list of
the valid cases, I opted to treat these cases identically.

- Raising an exception for "fatal."  It's possible that the dictionary
of valid cases is non-empty when "fatal" is encountered.  So it could be
argued that I should not raise an exception, but return what I've got.
I think that's misleading, and implies a canonical answer when that's
not necessarily the case.  Alternatively, the caller could wrap the ARC
lookup in a try/except clause, but I think that's optimizing for the
corner case.  If we start seeing fatal returns from the script, we can
always update it later.

- Alan: your note implied an extra field for the type of case, but the
script is not returning it.  I assume that's as expected, and the
"FastTrack" field in the sample output is not truly expected?

--Mark
_______________________________________________
tools-discuss mailing list
[email protected]

_______________________________________________
tools-discuss mailing list
[email protected]

Reply via email to