BIND 10 #963: Utility to upgrade SQLite schema
BIND 10 Development
do-not-reply at isc.org
Fri Mar 23 16:38:29 UTC 2012
#963: Utility to upgrade SQLite schema
-------------------------------------+-------------------------------------
Reporter: shane | Owner: vorner
Type: | Status: reviewing
enhancement | Milestone:
Priority: | Sprint-20120403
medium | Resolution:
Component: data | Sensitive: 0
source | Sub-Project: DNS
Keywords: | Estimated Difficulty: 0.0
Defect Severity: N/A | Total Hours: 0
Feature Depending on Ticket: |
sqlite schema upgrade |
Add Hours to Ticket: 0 |
Internal?: 0 |
-------------------------------------+-------------------------------------
Changes (by jelte):
* owner: jelte => vorner
Comment:
Ok, I have picked some specific issues from above, and randomly added one
or two of my own :)
- First off, I fixed the manpage creation, and added the manpage to the
repository (it's technically against my principles to do so, since it's
generated, but we do this for manpages so as not to depend on docbook etc
for normal compilation)
- I've slightly changed the run_dbutil.sh script to now change current
working directory, so it works with relative path for db file name
- I've replaced the direct system.exit values, and change --check to
return 1 on 'need-update', 2 on 'version too high for me', and higher for
other errors
- added 'to stdout' in the manpage verbose description
- modified output to use logging framework
- added -q --quiet option
And some other comments based on the discussion above:
- I think we can get away with not using the configuration db file name
for now; I think that if 324 (or any detection of out-of-date schema)
prints 'please run b10-dbutil --upgrade <database file>, it's good enough
for the near future
- Regarding different db types; renaming this tool was my first thought as
well. But having thought about it a bit, adding a --type option or
something seems better. If we don't take db info from existing
configuration I don't think we can autodetect; it just so happens that for
sqlite we have a file, but for most databases this will not be the case :)
But when we do, naturally we should default the --type to whatever happens
to be configured. And I think it would be better to keep these things
localized to as few tools as possible.
I also agree the tests should probably be done with lettuce; but that can
be deferred (in retrospect, I should've deferred the logging change too,
it's quite a big change, oh well).
did i miss anything else?
--
Ticket URL: <http://bind10.isc.org/ticket/963#comment:18>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list