bug in contrib/adddepend
Hi,
I just noticed a bug in adddepend:
The below commands will upgrade the foreign key style. Shall I execute
them?
DROP TRIGGER "RI_ConstraintTrigger_1105102" ON news_authors;
DROP TRIGGER "RI_ConstraintTrigger_1105103" ON news_authors;
DROP TRIGGER "RI_ConstraintTrigger_1105118" ON news_articles;
ALTER TABLE news_articles ADD CONSTRAINT "<unnamed>" FOREIGN
KEY (author)
REFERENCES news_authors(id) MATCH SIMPLE ON UPDATE NO
ACTION ON DELETE NO ACTION;
See how it's HTMLised the foreign key name? I cannot find how $keyname in
the code is being html escaped. Perhaps it's some weird taint mode thing?
Rod - you got any ideas?
Chris
On Thu, 2003-02-20 at 22:08, Christopher Kings-Lynne wrote:
Hi,
I just noticed a bug in adddepend:
The below commands will upgrade the foreign key style. Shall I execute
them?DROP TRIGGER "RI_ConstraintTrigger_1105102" ON news_authors;
DROP TRIGGER "RI_ConstraintTrigger_1105103" ON news_authors;
DROP TRIGGER "RI_ConstraintTrigger_1105118" ON news_articles;ALTER TABLE news_articles ADD CONSTRAINT "<unnamed>" FOREIGN
KEY (author)
REFERENCES news_authors(id) MATCH SIMPLE ON UPDATE NO
ACTION ON DELETE NO ACTION;See how it's HTMLised the foreign key name? I cannot find how $keyname in
the code is being html escaped. Perhaps it's some weird taint mode thing?
Not to mention when it's <unnamed>, the code is setup to ignore the
constraint name entirely (line 267).
There isn't anything that changes the value of $keyname after it's
pulled from the DBI drivers. What version of DBI::Pg are you using?
Last time I used it was with DBI::Pg v1.13.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
See how it's HTMLised the foreign key name? I cannot find how $keyname in
the code is being html escaped. Perhaps it's some weird taint mode thing?Not to mention when it's <unnamed>, the code is setup to ignore the
constraint name entirely (line 267).
I worked around it by checking to see if it equalled '<unnamed>' as
well as '<unnamed>'. I also have heaps of RI triggers left over after the
addepend run, so I'll investigate why adddepend doesn't pick them up...
There isn't anything that changes the value of $keyname after it's
pulled from the DBI drivers. What version of DBI::Pg are you using?Last time I used it was with DBI::Pg v1.13.
I'm using:
p5-DBD-Pg-1.21 Provides access to PostgreSQL databases through the
p5-DBI-1.28 The perl5 Database Interface. Required for DBD::*
Chris
I worked around it by checking to see if it equalled '<unnamed>' as
well as '<unnamed>'. I also have heaps of RI triggers left over after the
addepend run, so I'll investigate why adddepend doesn't pick them up...
Took a quick look, and I don't see anything in DBD-Pg that would do this
type of conversion.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
Was this resolved. Christopher, do you have a reproducible case?
---------------------------------------------------------------------------
Rod Taylor wrote:
-- Start of PGP signed section.
I worked around it by checking to see if it equalled '<unnamed>' as
well as '<unnamed>'. I also have heaps of RI triggers left over after the
addepend run, so I'll investigate why adddepend doesn't pick them up...Took a quick look, and I don't see anything in DBD-Pg that would do this
type of conversion.--
Rod Taylor <rbt@rbt.ca>PGP Key: http://www.rbt.ca/rbtpub.asc
-- End of PGP section, PGP failed!
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
On Thu, 2003-03-06 at 09:58, Bruce Momjian wrote:
Was this resolved. Christopher, do you have a reproducible case?
I've not been able to reproduce it, nor have I had a similar complaint.
I think it's a system configuration issue or somekind. Otherwise there
would have been many many more complaints, as <unknown> is a rather
common name for constraints :).
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
Was this resolved. Christopher, do you have a reproducible case?
It wasn't resolved, in fact I'd forgotten about it :)
I do have a reproducible case (our live server), however it seems like it's
basically a case of an invalid set of triggers. I really need to manually
remove some of the triggers. eg. They're supporting one side of the foreign
key, but not the other, etc...
Chris
Was this resolved. Christopher, do you have a reproducible case?
Oh sorry, I answered the wrong question!
Yes, I resolved it by reinstalling my DBD perl stuff.
I still have the problem of left over constraint triggers, but they do look
like they're broken, so it's not an adddepend problem...
Chris