Zeoslib Portal

Congratulations! )))
The function could be deprecated but many people tend to stick to older versions (I've seen pretty much guys still using Firebird 1.5!) so in Zeos we'll have to support even the most ancient pieces of a mammoth's sh*t.

I've little knowledge on MySQL but pid hardly could be 0, I guess you should first get the right value with mysql_thread_id() (according to docs). Moreover the function returns a result, maybe it could shed some light.

Yep. FB has cancelling as well though I've never used it. As for the fact the functions were commented out, I guess nobody have touched them until this time :mrgreen: . I think this should be a general method implemented for all drivers just like PingServer. But for now you can implement the method ...

Suspending a thread from outside is deprecated and non-reliable. I'm sure this kind of stuff should not go to prod/release versions.

Well, cancelling a current operation should exist in some form for all drivers. Cancelling a connect is more complicated though... Anyway I don't think you'll achieve what you want by unloading libs. There could exist some internally allocated variables which won't be disposed on thread termination ...

Ah, I thought you're talking about threaded BLOB loading only :) Well, terminating a thread is always a very bad practice and should be used only as extreme measure. I'd recommend trying API functions for that, f.ex. for MySQL functions like mysql_kill() / mysql_query() with KILL stmt; connect-stage...

When you call WinAPI TerminateThread, you just halt the Execute method but nothing prevents thread destructor or any other method from running. So just dispose all objects you created there.

CodeTyphon is nothing more than pre-configured Lazarus in fact

Hello Fr0st, you are absolutely right. I am just not sure, if we have all the support in Zeos we would need for that: - A GenericCachedResolver, that generates statemens using the returning clause, when possible - a way for the generic resolver to get the changed values into a (Cached)ResultSet ups...

Isn't that possible with RETURNING clause?

Pretty nice idea! Some more ideas: - call event with currentrow/rowcount parameters - add Cancel flag as well (though it should be carefully checked!) - add a field / option to fire only on N-th iteration (nevertheless, I doubt that firing for every row will slow things down too much for usual cases...

Wow... I wish everyone reported issues so thoroughly! My respects

I guess where you said 1251 should be 1250 cause 1251 is Cyrillic that doesn't have Polish symbols. But anyway check again you haven't confused them

Adaptation to Nextgen compiler is in progress, you can see numerous IFDEF NEXTGEN here and there

From what you said, you have your "boolean" char(1) without any explicit codepage. Hence it's utf8 too. One Utf8 char could occupy up to 4 bytes. Even in FB itself the maximum length of UTF8 VARCHAR is 8k chars while ANSI encoded field could take 32k. So the field size is correct in terms ...