Bugzilla

It may be related to this:

(D. Richard Hipp from bug 1614967)
> In order to support APIs like [sqlite3_filename_database()](https://www.sqlite.org/draft/c3ref/filename_database.html), and to support API calls like [sqlite3_uri_boolean()](https://www.sqlite.org/draft/c3ref/uri_boolean.html) against the names of journal files passed down into an SQLite VFS, the filenames must be laid out in memory in a very particular way.  Older versions of SQLite were more tolerant of non-conforming memory layouts.  Newer versions are increasingly strict.  My guess (and this is only a guess) is that the TelemetryVFS does not conform to the latest (evolving) requirements on the memory layout of filenames passed into xOpen.
> 
> There is a new API for SQLite 3.32.0 (due for release on 2020-05-29) that will construct filenames suitable to be passed into the xOpen method of a VFS.  See draft documentation at <https://www.sqlite.org/draft/c3ref/create_filename.html>
> 
> Long-term, I think the correct fix is to update TelemetryVFS to make use of sqlite3_create_filename() when it constructs filenames to be passed down into lower-level VFSes.  That API is not yet available in an official SQLite release, but the [Prerelease Snapshot](https://sqlite.org/download.html) is certainly stable enough for development work, and if you ask, we can run tests to ensure that it is stable enough for an official FF release.

It may be related to this:

(D. Richard Hipp from bug 1614967)
> In order to support APIs like [sqlite3_filename_database()](https://www.sqlite.org/draft/c3ref/filename_database.html), and to support API calls like [sqlite3_uri_boolean()](https://www.sqlite.org/draft/c3ref/uri_boolean.html) against the names of journal files passed down into an SQLite VFS, the filenames must be laid out in memory in a very particular way.  Older versions of SQLite were more tolerant of non-conforming memory layouts.  Newer versions are increasingly strict.
> 
> There is a new API for SQLite 3.32.0 (due for release on 2020-05-29) that will construct filenames suitable to be passed into the xOpen method of a VFS.  See draft documentation at <https://www.sqlite.org/draft/c3ref/create_filename.html>
> 
> Long-term, I think the correct fix is to update TelemetryVFS to make use of sqlite3_create_filename() when it constructs filenames to be passed down into lower-level VFSes.  That API is not yet available in an official SQLite release, but the [Prerelease Snapshot](https://sqlite.org/download.html) is certainly stable enough for development work, and if you ask, we can run tests to ensure that it is stable enough for an official FF release.

It may be related to this:

(D. Richard Hipp from bug 1614967)
> In order to support APIs like [sqlite3_filename_database()](https://www.sqlite.org/draft/c3ref/filename_database.html), and to support API calls like [sqlite3_uri_boolean()](https://www.sqlite.org/draft/c3ref/uri_boolean.html) against the names of journal files passed down into an SQLite VFS, the filenames must be laid out in memory in a very particular way.  Older versions of SQLite were more tolerant of non-conforming memory layouts.  Newer versions are increasingly strict. [...]
> 
> There is a new API for SQLite 3.32.0 (due for release on 2020-05-29) that will construct filenames suitable to be passed into the xOpen method of a VFS.  See draft documentation at <https://www.sqlite.org/draft/c3ref/create_filename.html>
> 
> Long-term, I think the correct fix is to update TelemetryVFS to make use of sqlite3_create_filename() when it constructs filenames to be passed down into lower-level VFSes.  That API is not yet available in an official SQLite release, but the [Prerelease Snapshot](https://sqlite.org/download.html) is certainly stable enough for development work, and if you ask, we can run tests to ensure that it is stable enough for an official FF release.

It may be related to this:

(D. Richard Hipp from bug 1614967)
> In order to support APIs like [sqlite3_filename_database()](https://www.sqlite.org/draft/c3ref/filename_database.html), and to support API calls like [sqlite3_uri_boolean()](https://www.sqlite.org/draft/c3ref/uri_boolean.html) against the names of journal files passed down into an SQLite VFS, the filenames must be laid out in memory in a very particular way.  Older versions of SQLite were more tolerant of non-conforming memory layouts.  Newer versions are increasingly strict. [...]
> 
> There is a new API for SQLite 3.32.0 (due for release on 2020-05-29) that will construct filenames suitable to be passed into the xOpen method of a VFS.  See draft documentation at <https://www.sqlite.org/draft/c3ref/create_filename.html>
> 
> [...]update TelemetryVFS to make use of sqlite3_create_filename() when it constructs filenames to be passed down into lower-level VFSes.  That API is not yet available in an official SQLite release, but the [Prerelease Snapshot](https://sqlite.org/download.html) is certainly stable enough for development work, and if you ask, we can run tests to ensure that it is stable enough for an official FF release.

nvm, I misread that.