Hi,
On Tue, 2009-03-31 at 21:37 +0200, Adeodato Simó wrote:
> Package: libsqlite3-0
> Version: 3.6.11-4
> Severity: serious
>
> The sqlite3_io_methods struct, which just contains pointers to different
> functions, changed in an incompatible way, by making one of these
> methods (xCheckReservedLock) accept two parameters instead of one.
[...]
> It is unclear to us what is SQLite’s compromises regarding this API,
> since it is true it’s a very low-level functionality, and possibly the
> SQLite authors are expecting for code using it to be compiled together
> with SQLite itself. But, alas, it gets exported via sqlite.h, so...
What I have found[1] states:
"SQLite version 3.6.0 contains many changes. As is the custom with the
SQLite project, most changes are fully backwards compatible. However, a
few of the changes in version 3.6.0 are incompatible and might require
modifications to application code and/or makefiles. This document is a
briefing on the changes in SQLite 3.6.0 with special attention to the
incompatible changes.
Key Points:
* The database file format is unchanged.
* All incompatibilities are on obscure interfaces and
hence should have zero impact on most applications."
[...]
"The signature of the xCheckReservedLock method on sqlite3_io_methods
has been modified so that it returns an error code and stores its
boolean result into an integer pointed to by a parameter. In association
with this change, a new extended error code
SQLITE_IOERR_CHECKRESERVEDLOCK has been added."
The keyword seems to be obscure interfaces. Also the extra argument of
xCheckReservedLock indicates that the underlying method is not changed
too much. It's only return that error code defined as a parameter.
> We are going to rebuild evolution-data-server in unstable as to give a
> temporary fix to the issue. We don’t know if other applications are
> affected, and we should assess what permanent solution to give to this
> problem.
A local rebuild of evolution-data-server and evolution itself was
built; but not tested. BinNMU seems to be sufficient for this case.
Regards,
Laszlo/GCS
[1] http://www.sqlite.org/35to36.html
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
