[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [sqlite] PRAGMA bug in 3.7.7 (but fine in 3.7.6.3)

From: Greg Stein <gstein_at_gmail.com>
Date: Sun, 26 Jun 2011 19:20:30 -0400

On Sun, Jun 26, 2011 at 18:52, Richard Hipp <drh_at_sqlite.org> wrote:
> On Sun, Jun 26, 2011 at 3:59 PM, Greg Stein <gstein_at_gmail.com> wrote:
>
>> Hey all,
>>
>> There is a bug in 3.7.7 that affects Apache Subversion. We've verified
>> that it is this latest release, as prior releases have worked just
>> fine for us.
>>
>> Below is the reproduction script, thanks to Daniel Shahaf. With this
>> pragma not working, we cannot even start up :-(
>>
>
> Thanks for the test case.
>
> The pragma is actually working.  The bug is that it is returning
> SQLITE_SCHEMA instead of SQLITE_OK.  If you wanted to, you could just ignore
> the error and keep going and everything would work.

Okay. I'll add some code to ignore the result *if* we have 3.7.7 *and*
it is SQLITE_SCHEMA. Thus, we can keep trapping other issues and on
other releases.

>
> The bug was introduced here:
>
>    http://www.sqlite.org/src/ci/957b2ab67c
>
> And more specifically with the removal of the "p->expired = 0" line which is
> shown as the very first change here:
>
>    http://www.sqlite.org/src/fdiff?v1=0505dc4f7ff3cf35&v2=0b2e2880f13af400

Yup. "Whoops" :-)

[ also: there is a typo in the docstring for
sqlite3VdbeAddParseSchemaOp. s/local/lock/ ]

> The big mystery is why this problem is not showing up in our rather
> extensive test suite.  I'm still investigating that aspect of the issue.

Yeah. I was pretty surprised because I knew how much testing goes on
with SQLite. Looks like the test case count just grew by one :-)

Cheers,
-g
Received on 2011-06-27 01:21:00 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.