mORMot Open Source

zero is invalid, a valid ID value is >=1.

Under certain conditions the internal encoding of a TRestBatch.add action changes.

Problem can be traced to mormot.orm.sqlite3, specifically to procedure TRestOrmServerDB.InternalBatchStop, specifically lines 2615-2628.
When the encoding is in BATCH_DIRECT_ADD it makes a call to EncodeMultiInsertSQLite3 with the SimpleFields parameter. Because TRecordVersion is not a "SimpleField" it's not accounted for and as such a discrepancy occurs between number of bits in SimpleFields (n) and actual fieldCount (n + 1) getting passed to this procedure.

Anyway, that's all from me, don't have the time right now but I hope this helps a bit.

that's how you're supposed to use it.  The only reason you can still call Add/Update/Delete directly from the TRestServerDB instance is because it's there for backwards compatibility with mORMot 1. If you define PUREMORMOT2 flag for your project then you will see those direct methods no longer available.

By default there is a lock at orm level so all threads orm access will be properly synchronized, shouldn't have a problem.

I'm in same boat as sakura, basically I've switched to mORMot 2 during development but aren't using it in production yet because the project isn't there yet.

Interesting, I personally use "DB Browser for Sqlite" also known as sqlitebrowser, open source tool https://github.com/sqlitebrowser/sqlitebrowser

tbh the no root is better to be handled by whatever front-end reverse/balancing proxy you are using. It's trivial to do this in nginx.

AFAIK it's already documented on how to call interface based services on server side : https://synopse.info/files/html/Synopse … #TITLE_421

The issue you have is that you're executing database requests in the same thread as the user UI which is just bad all around. You should decouple your logic from the UI. Database requests should be done in a separate thread. However, this requires some threading knowledge on your own part. Please read : https://synopse.info/files/html/Synopse … #TITLE_304

Correct me if I'm wrong, but it seems from your comments that your primary concern is the user UI stalling/locking/delaying due to waiting for database action to complete?

Nice that you found out why but it really isn't the tools fault, there was clearly a conflict with your manual changes. The tool did it's job and prevented the loss of your modifications, it warned you that there is a conflict but you probably just closed the window. What you had to do is to click on the conflicted file in the update window and click manual resolve.... Then you would get an editor where you could manually merge or update the conflicted lines. The same thing would happen if you used any other tool/client.

I also like the Tortoise client, the only thing I change is that I use Beyond Compare as the editor of choice for merges&conflicts, highly recommend.

Oh, didn't know you are using postgre, I have no experience with that database, maybe someone else will chime in, gl.

Sounds like he needs a pivot table instead.

Not sure where you got that date from but 3.2.0 was released in 2020. Honestly if you're having issues just download 3.2.0 from offical freepascal webpage.

mORMot links external object files that haven't been compiled with -fPIC, you can avoid the errors by using external libraries instead,
for the errors you provided you need to set the following compile time flags :

NOSQLITE3STATIC
NOLIBDEFLATESTATIC

You can do this on command line or if IDE , for example lazarus ide :
Project Options -> Compiler Options -> Custom Options

I've done this for a different reason then you and I don't recommend it unless you are a seasoned veteran of writing external libraries. Things like shared memory will be prohibited because the framework wasn't designed with relocatable code in mind. mORMot has a lot of initialization hooks that hook into exceptions, rtti and system units and while there are (some) compile time flags that control this behavior it really goes against the "grain" of the project and may result in unwanted behavior.

That much said, if you don't care about shared memory, mORMot will work just fine in a library after all, without shared memory there's little difference between a program and a library.
For your use case I fail to see why you would want to use a library instead of a normal program, everything you mentioned can be achieved with writing a normal program and in your case there's only disadvantages to using a library instead.

Hard to tell unless you provide a minimal example we can test, tbh I suspect it might be the new httpasync server which is the default in mORMot 2, can you test with the normal http socket server, or if you're using websockets, BidirSocket server.

TRestHttpServer.Create(..., TRestHttpServerUse.useHttpSocket,..l);

hey, the memory numbers you see in the process tab of the task manager are just a crude approximation of actual memory used. If you have a lot of heap de/allocations like in your case then you are bound to have some heap fragmentation. This means the heap memory will grow over time (but should stabilize) which is what you are observing. There's not much wrong with that since the memory manager will try to reuse the fragmented space.

Absolutely the better way is to use a memory profiler instead of the task manager. If you have to use external software then the better choice is Microsoft's Process Explorer from SysInternals (You would look at the "Private Bytes" statistic of Virtual Memory), but again, it's all just crude approximation.

thanks, that works!

Nice, I didn't even know someone was creating deb packages (I heard it's PITA), I think both ways should be documented. Also for the packages, a hash should be posted in the documentation, for security.

Initialization is not the problem, in my project I have everything set up correctly with constructor&destructor in a try..finally block and still produces this bug, I also tested it on this MWE example.
It's not that big of a deal tho because nobody should be calling ItemPtr function in production anyway since you can access the original Array directly, however it would be nice if TSynDictionary would have a public function that gives you a pointer to the start of the Keys & Values array respectively.

Anyone here who likes to modularize their applications (linux) by writing dynamic libraries? If so, do you use shared memory manager and which one? Any pitfalls and recommendations ?

VPS or dedicated?