On Wed, 2012-06-06 at 10:50 +0600, Timon wrote:
> seems that this commit broke reindexing of selected index. steps to reproduce:
> 1) create table
> 2) create index
> 3) select index in object inspector
> 4) try to reindex it via maintenance menu item
> 5) got error : ERROR: schema "table_name" does not exist
> and one more crash here
> .. same steps as before
> 4) try to CLUSTER index
> 5) pgadmin simply crashed
OK, I finally got some time to work on this. As Timon said, these bugs
come from the patch "Lots of work on domains, and check constraints". In
this patch, I changed some objects parent class from pgTableObject to
pgSchemaObject. Due to this change, the GetTable() method returns NULL,
which segfaults all statements that try to use the return value without
checking. The two examples above from Timon are exactly this.
I don't see many ways to get out of this issue.
We could use GetSchema() instead of GetTable(). It works, it's an easy
and small patch. But it'll certainly be a maintenance nightmare (at
least without any comments)
We could also revert my patch. It's simple, we loose the feature of
adding as many check constraints as we want to a domain, we loose the
feature of renaming and validating constraints, and we gain a few bugs.
I don't see any other options. My own personal choice would be the first
one (see attached patch). But it's a tough call.
In response to
pgadmin-hackers by date
|Next:||From: Guillaume Lelarge||Date: 2012-07-21 13:26:38|
|Subject: pgAdmin III commit: Add *.dirstamp pattern to .gitignore|
|Previous:||From: pgAdmin Trac||Date: 2012-07-20 17:04:14|
|Subject: Re: [pgAdmin III] #371: Enhance the search tool|