valentina 2 addRecord by API

[Olivier]

I continued my tests by trying the addRecord of the API. The first time 
it was with a cursor.
It is faster but it is slow even MUCH MORE than with Valentina 1.
The tests are made with at most 36000 simple lines of only 4 small 
fields without links.
But with a lot more lines and complex recordings...

Example :
AddRecord of records of 4 fields (3 strings, 1 boolean). The 3 strings 
(lenght 38) are indexed. one unique. one indexedByWord.

Valentina 1 : cache 20 Mo.
Valentina 2 : cache 50 Mo. I tried 20 or 90 Mb, it is the same thing.
RB 5.5.4

Add x records	Valentina 1	Valentina 2 CURSOR	Valentina 2 API

Add 100		2 secondes		4 secondes		3 secondes
Add 600		2 sec			11 sec			7 sec
Add 3000		3 sec			50 sec			26 sec
Add 5000		3 sec			86 sec			46 sec
Add 8000		4 sec			147 sec			83 sec
Add 11000	5 sec			212 sec			119 sec
Add 17000	6 sec			355 sec			211 sec
Add 25000	8 sec			547 sec			329 sec
Add 36000	10 sec			875 sec			542 sec

Even if three strings are NOT indexed and unique,
on the addrecord of 8000 lines, Valentina 2 puts 114 seconds (cursor) 
or 54 seconds (API). The gain is modest.

Valentina 2 is UTF16 but in Valentina 1 EVERY string is TRANSFORM of 
UTF8 in ASCII-FRENCH by a RB METHOD before the recording in DB.

Le 8 janv. 05, à 07:47, Olivier a écrit :

> Hi Ruslan and list,
>
>
>> Right now no. I see that 2.0 now is 2 times slower on addRecord.
>> Don't know about rest operations.
>> But we yet in debug mode for indexes for example.
>>
>> I think that AFTER we spend a month or so on optimzation and tunning,
>> We will make 2.0 faster of 1.x.
>
> Effectively, addRecord is MUCH slower. Far too much.
>
> Example :
> AddRecord of records of 4 fields (3 strings, 1 boolean). The 3 strings 
> (lenght 38) are indexed. one unique. one indexedByWord.
>
> Valentina 1 : cache 20 Mo.
> Valentina 2 : cache 50 Mo.
> RB 5.5.4
>
> Add x records		Valentina 1		Valentina 2
>
> Add 100			2 secondes		4 secondes
> Add 600			2 sec			11 sec
> Add 3000			3 sec			50 sec
> Add 5000			3 sec			86 sec
> Add 8000			4 sec			147 sec
> Add 11000		5 sec			212 sec
> Add 17000		6 sec			355 sec
> Add 25000		8 sec			547 sec
> Add 36000		10 sec			875 sec
>
> Even if three strings are NOT indexed and unique,
> on the addrecord of 8000 lines, Valentina 2 puts 114 seconds. The gain 
> is modest.
>
> Valentina 2 is UTF16 but in Valentina 1 EVERY string is TRANSFORM of 
> UTF8 in ASCII-FRENCH by a RB METHOD.
>
> Are you SURE that you can optimize and accelerate significantly 
> Valentina 2.0 ?
> That you will reach at least the same speed as Valentina 1?
> Because the distance is really very important.
>
> thank you
>
> olivier
>
> _______________________________________________
> Valentina-beta mailing list
> Valentina-beta at lists.macserve.net
> http://lists.macserve.net/mailman/listinfo/valentina-beta
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 3379 bytes
Desc: not available
Url : http://lists.macserve.net/pipermail/valentina-beta/attachments/20050108/7c68a47f/attachment.bin