Hi!
Øystein Grøvlen wrote:
> Changes look good. Test succeeds with fix and fails without.
>
> I have two questions:
>
> - In Bug#33567, Peter says that "users wouldn't enter the same name
> twice except in error, so this should be an error." Have you
> discussed your solution (to just ignore duplicates) with Peter?
I have checked with Peter and he thinks the statement should error, as other
statements do in a similar situation. I have changed the patch to implement this
behaviour. I do it as for other statements, i.e. do check for duplicates inside
the parser.
Thus, in the new patch, parser will detect and report duplicate database names
as errors. But backup kernel remains as before: if it ever gets the same
database twice in the list, it will ignore the second occurrence and never try
to backup the same database twice.
>
> - Is it difficult to test that the backup image is no longer
> larger with duplication as reported in BUG#34480?
>
What would you like to compare against? BACKUP is fixed now and never produces
images with duplication. I think Ingo's backup client program will be helpful in
testing this and a lot of other things. When the client is ready we should add
plenty of tests checking that the image contains what we expect it to contain.
Please check your review box if you think the new patch is good for pushing.
Rafal
