[Solved] Fast Real-Time File Read Write

[argumentum]

you can, for now, copy it to your script folder and load it by _SQLite_Startup(@ScriptDir & "\sqlite3.dll") instead of _SQLite_Startup() .

[argumentum]

...I can't make it work for myself now. I'll see wassup with that and get back to you 😓
, tho, if you got it working, never mind me.

[argumentum]

lol, read the manual 
_SQLite_Startup(@ScriptDir & '\sqlite3.dll', False, 1) will do it  

[Ascer]

It works for me now.

_SQLite_LibVersion=3.31.1
Update of row natation took 7.03948001411146 ms (dik)
Update of all rows where login start with 'seq' took 54.5459210399718 ms (disk)
Copy users table (109k rows + index) to memory took 109.470713377022 ms
Update of row natation took 0.513999208903048 ms (memory)
Update of all rows where login start with 'seq' took 30.038165082691 ms (memory)
Copy back changes to disk took 12.007141253568 ms

Now i need to spent a more time to understood how this all functions works and if they fit to this job or not.

[jchd]

1 hour ago, Ascer said:

In example below i used 1000 users cuz creation of 100,000 take to long and required 1,12GB pc space

You seem to create the table inside the loop!

Also inserting every row by itself is very slow; wrap the whole loop inside a transaction.

1.12 Gb for 100k rows is enormous (about 11.2 kb per row ???). My users.sq3 example also has ~100k rows but is only a 11 Mb file...

[TheSaint]

Just for the record, yes I am getting older .... all the time in fact. 

Not sure you can apply fat and lazy to me though, just because I prefer to usually give my coding help in other ways ... something I have always done ... prior to becoming MVP too. 

Take all the programs and scripts I regularly provide here or update, all for free ... no one makes me do it, and hardly something a lazy person would bother with. 

Here is a slightly related one to this topic ... there are others.

And here is a doorway to many more, if mostly unrelated.

And I have provided many many more.

On behalf of all good fat people everywhere, I take offense. 

@Jos - Don't worry, this is not the start of an ongoing conversation, just illumination for those who need it. I will now leave this topic in peace.

[Ascer]

@Jos @RTFC @jchd @argumentum @Nine @TheSaint @LarsJ @condoman

Thanks everyone for help.

After a few days of tests where I used many different options i decided to stay with Dictionary. Edited main thread with solution for ppls with this same problem.

Below speed results for using 3 methods to find user nr 100,000 where his table was {name, passwd, range, group}.

• [ADODB.Connection]:  8ms real-time save/write to file
• [_SQLite_]:  21ms real-time save/write to file
• [Dictionary+XML]: 0.06ms memory edit, to save file need spent additional 2s.