Zeoslib Portal

marsupilami wrote: ↑16.07.2023, 12:19 On the DBC layer the TZGenericResolver does exactly that. It looks for a table name and then queries for the primary key.

But how it looks for table name? And how it could know when key field is aliased?
Anyway there are VIEWS that could contain literally everything

The most correct way is 2nd independent transaction. IDK about Postgres, maybe it would require second connection.

I solved my issue by imported and exported keys in the mean time which seem to be doing the trick for my purpose. However it would be really nice if we could get TField.IsIndexField working under Zeos :) Not reliable in general... A field in a query could be taken from any table, aliased, taken fro...

However an other suspicious line in TryUniToDate: else if (F in [Byte('-'),Byte('/'),Byte('\'),Byte(' ')]) then begin Shouldn't we be using FormatSettings.DateSeparator instead of a constant array...? I guess TryUniToTimeStamp is meant to be a kind of universal converter regardless of some regional...

In fact, everything that compiles in Delphi 7 compiles without problems in subsequent non-Unicode Delphi versions (not always the other way round). Unless you don't have all the sources, then there's a problem. Fair point. I revised the list of updates in 2005 and 2006 and it includes quite useful ...

Yes it compiles but I was worrying about the errors you reported.
Upgrading is the option I wish to avoid - the app is not worth the efforts. So far I have no issues with D7-built apps on W10

Version is better to be 2-component: major and minor; minor versions are compatible at least backwards (newer software could read older data). However, in my opinion the format that is supposed for massive interchange should be more standard, probably XML, MsgPack or ProtoBuf. Regarding header: exte...

As for the D7, I wouldn't recommend Zeos8 for it, I get too many unexpected errors, it is true that testing on Windows11 (maybe too much hardcore). Ouch, bad news. I was going to upgrade my D7 app to the fresh Zeos. The problem is that it used trunk 7.3 which has never been released and turned to 8...

marsupilami wrote: ↑12.01.2023, 09:40 We have a hard coded default port for MySQL? I agree it should be removed but I currently cannto spare the time to do that. Couls you maybe prepare a patch?

Turns out we do
Sure, this is really tiny patch but want to ensure it builds so I'll do it on Monday

More to say, Zeos uses no default ports at all (except for MySQL - and I think it should be removed - https://dev.mysql.com/doc/c-api/8.0/en/mysql-real-connect.html) leaving this option to user or driver internals (config/default). And I think it's right. Transport level should be left to the driver...

Maybe GetLoadedLibrary?

Well, common approach to solving almost any issue is localize it first. So trace all the lines up to DB client lib call and push checks here and there (luckily Zeos comes with sources so you can modify the code). Another way is to find the exact Zeos commit which introduced the issue. If you say you...

Heisenbug :( Are you building with Delphi? Try to include FastMM4 and turn on all memory corruption detection options. Maybe this will give some hint BTW. Your dumps seem to show that BLOBs contain all source contents plus a tail of (previous?) source contents while in the initial post you complaine...

Well, that's SQLite so I guess the DB is local - some actions could be sent to background threads.

P.S.

OMG, this is the most paranoid code I've ever seen... Anyway it really doesn't check much. If you really are to check the loading, save blob to another memory stream and compare it with source one. Moreover, here you're checking only local assignment. Why not check if a blob was correctly saved to D...