Reindex Success?

[Ruslan Zasukhin]

On 6/10/04 11:11 PM, "Keith DeLong" <delong at redcort.com> wrote:

>> You know, it looks you have raise good question.
>> 
>> Right now it is not possible to see if it was done fine.
>> 
>> Btw, in 2.0  we will get working diagnose() for indexes!
>> This will allow us open eyes: if index is good now.
> 
> I'm looking forward to 2.0 and I understand it is properly your focus.
> 
> As I've mentioned before, I'll be shipping apps with version 1 for at least
> 6-9 months after 2.0 is finally released. I know you've disagreed, but
> there's so much new I will need to let version 2 mature and stabilize many
> months before rolling out a commercial release embedded with version 2.

I agree. But because in any case adoption will take time.

About maturing...you can see that in 2.0 we close big holes -- i.e. Semantic
bugs that was in Valentina 1.x.

E.g. This assignment to ObjectPtr of not-existed RecID.
Just we all take RecID from table, so it must be there.
But safety is safety.
 
And we discover such holes thanks to huge number of tests that we prepare
here. I think 50% of our work and code now -- this is tests.

Another example.  In Valentina 1.x, I have told many time on list:
    if you need 1:1 link, just make ObjectPtr field unique.

We have made tests on this, thinking, hey, what here can be hard.
But ops. Not works. And deep analyze have found that we need CHANGE semantic
of ObjectPtr field !!!

- ObjectPtr field will be nullable as any other field in 2.0.
- ObjectPtr field in 99% must be created as nullable.
- value ZERO for ObjectPtr will be PROHIBITED, because in T1 not exists
record with RecID = ZERO.
- to find records with no parent we must write now
    not     WHERE ptr = 0
    but     WHERE ptr is NULL.

And all this discovery, only thank that we are not lazy, and we try now TEST
any and each aspect and feature of engine.

This make me think that Valentina 2.0 from born will be very solid.
Of course some issues can be found only in real life.

And we have not yet even start develop similar huge system of benchmark
tests, I mean such functions that are specially designed to help profile
engine. 

> Because of this I would be very pleased if you'd make a short list of 1.10
> fixes to be done at some point. I'd gladly pay a v1 upgrade fee to see
> version 1 continue to be supported and fixed as appropriate for the next
> year to 18 months.

Well, I think we will contain keep 1.x some time on FTP.
Support will be not urgent. And from other side not so many bug reports we
have for 1.x.

And as I know right now we do not have show-stopper bugs.


-- 
Best regards,
Ruslan Zasukhin      [ I feel the need...the need for speed ]
-------------------------------------------------------------
e-mail: ruslan at paradigmasoft.com
web: http://www.paradigmasoft.com

To subscribe to the Valentina mail list go to:
http://lists.macserve.net/mailman/listinfo/valentina
-------------------------------------------------------------