Notice (2018-05-24): bugzilla.xamarin.com is now in
Please join us on
Visual Studio Developer Community and in the
Mono organizations on
GitHub to continue tracking issues. Bugzilla will remain
available for reference in read-only mode. We will continue to work
on open Bugzilla bugs, copy them to the new locations
as needed for follow-up, and add the new items under Related
Our sincere thanks to everyone who has contributed on this bug
tracker over the years. Thanks also for your understanding as we
make these adjustments and improvements for the future.
Please create a new report on
Developer Community or GitHub with
your current version information, steps to reproduce, and relevant error
messages or log files if you are hitting an issue that looks similar to
this resolved bug and you do not yet see a matching new report.
# Steps to reproduce
Download the sample code from - https://github.com/tomhamming/xamarin-bundle-sqlite
Open a database connection and configure it as WAL journaling with a 10 second (or so) transaction timeout.
Start a transaction and do some nontrivial work (simulated as a sleep in this case)
On another thread, open another connection with the same configuration to the same DB file. Open a transaction.
# Expected behavior
Expected behaviour is that the second transaction should block until the first one finishes, or return SQLITE_BUSY if the first transaction lasts longer than the second one's timeout value. This is what happens under Xamarin on the simulator, and under Objective-C on both the simulator and device.
But under Xamarin on the device, the second transaction fails with SQLITE_BUSY immediately
# Actual behavior
Second transaction should wait for the first transaction to complete
# Supplemental info (logs, images, videos)
Apple does not use the default build settings for sqlite3, including different default for concurrency http://sqlite.org/threadsafe.html - that should not be an issue in this case (since you're not looking to serialize the calls) but it can explain difference you get with the simulator (and other sqlite environments).
If you want to use a custom libsqlite (linked staticaly) then _all_ `DllImport` must be set to __Internal (where the code will end up) and at least one is missing the linked test case.
> [DllImport("sqlite3", CallingConvention = CallingConvention.Cdecl, CharSet = CharSet.Unicode)]
^ you'll end up using two different libraries (and that won't work).
I've tried the __Internal DllImport statements and they don't work. The current revision of the example repo is configured to use the included sqlite library, which is at version 3.9.2, but it is still linking against the sqlite in iOS (220.127.116.11).
Correction: *now* all the DllImport statements are __Internal. But it's still not statically linking against the included sqlite file.
@Rolf I'm sure that used to work but maybe recent changes affected how we deal with libsqlite3 ?
It's notable that under iOS 9.2, the issue manifests both on the device and simulator, including when running from Xcode, when linking against the iOS sqlite library (under 9.1, it was only a problem running on a device from Xamarin). Everything works fine from Xcode if I bundle my own sqlite in the Xcode project. So perhaps the workaround here is to get Xamarin.iOS to link against a bundled sqlite version.
One problem is that the symbols in libot_sqlite.a are private:
> nm -m libot_sqlite.a | grep sqlite3_libversion
000000000000d1bb (__TEXT,__text) private external _sqlite3_libversion
and that won't work with Xamarin.iOS. Since symbols are looked up dynamically (using dlsym, which won't find private symbols), Xamarin.iOS requires symbols to be public.
@Tom, can you try rebuilding sqlite with public symbols and see if that works?
@Rolf hot dog, that did it! For reference, the build setting in Xcode is called "Symbols Hidden by Default" under "Apple LLVM - Code Generation". It might be worthwhile to put that on the page about linking native libraries (https://developer.xamarin.com/guides/ios/advanced_topics/native_interop/). That would have saved me a lot of time.
Thanks for looking at this.
Great! I'll mark this as resolved then.
I'll also see if I can update the documentation.