Zeoslib Portal

But maybe that's not needed. Afaik it's unnecessary, as GetInstance returns a new instance. As long as there's a plain driver, you'll always get an instance. I did not check the code right now, though so I might be wrong. If anyone is about to add this please make sure that it will continue to work...

I am using it in production since it was still called 7.3 without any major roadblocks. Demos should be the same for Lazarus and Delphi... if you tell us what you would like to do, maybe someone can point you to the demo you are looking for. As for the TZTransaction I can not give an answer as I per...

It would also be nice to be able to determine which library is currently being used to see if the correct one is in use. It's actually possible, I even posted it on this forum a while ago. Can not find it atm, so here you go: If Assigned(ZConnection1.DbcConnection) And Assigned(ZConnection1.DbcConn...

Probably this is going to be a driver issue, as opening the exact same resultset on the exact same machine using the exact same settings, but using OCI 19.8 loads everything perfectly. The new version also loads all data correctly, if LobCacheMode is None. I just would like to leave it here as it se...

Can you check the metadata, if NULLABLE is properly returned and detected by Zeos?
Also, a minimal example would be appreciated so we can check what is going wrong.

Maybe you don't need to interrupt the network. Check if killing the session is sufficient, this would make testing a lot more easy: https://dataedo.com/kb/query/mysql/kill-session MySQL's current implementation of .AbortOperation is similar: function TZMySQLConnection.AbortOperation: Integer; Var ki...

I saw that Michael took a different approach fixing this, the commit is already in SubVersion . However I don't see the reason for a required check to be implemented 3 different locations (once in TDataSet, once in TZAbstractRODataset and now in TZMemTableResultSet) this solution feels a lot more el...

Does anyone know what this means by the way? This is why inherited isn't called, this is why required fields are simply ignored. //AVZ - Firebird defaults come through when this is commented out Maybe it worth a try to simply call inherited and run it against the tests. I would like that solution be...

I have no SF account so can not post a comment on the bug. https://sourceforge.net/p/zeoslib/tickets/560/ 1, blob.Clear seems to work as intended. ZMemTable1.FieldDefs.Add('ID', ftLargeint, 0, True); ZMemTable1.FieldDefs.Add('Name', ftString, 255, True); ZMemTable1.FieldDefs.Add('Data', ftBlob); ZMe...

Fixed :)

Pull request available on GitHub.

The latest Zeos8 from SVN or Git works just fine.
If you prefer, you also can try to install 7.2 as well with the latest package available; it might work.

OMG, this is the most paranoid code I've ever seen... "World state" suggests we are talking about a game engine. Imagine the lag 1,2 sec of saving world state will generate :) Tbh I don't see the reason of this check either. I'd only do safeguards if actual corruption happened once but co...

This by itself won't do much. - Save the original data to the disk, save the contents of blob field to the disk. Compare the differences. - After saving your blob, post the dataset and read it to a different stream. Save both streams to the disk and compare differences. My guess is your data is safe...

Fresh, ripe pull request available for picking! :)

You'll need something like TZGenericStatementAnalyser and DefineSelectSchemaFromQuery but instead of SELECT and FROM, you'll have to search to DELETE FROM, INSERT INTO and UPDATE SET. Most probably you'll end up with a class helper / derived analyzer. I just wrote one for myself, that's why I know ...