Wow, I'm really impressed with the support of this task - thank you. Although it sounds like I'm being unemployed  But I will for sure prefer if we can make a general solution.

I am a little bit unsure how this will work in practice. it sounds like a switch/attribute we can set, if we want another way of sorting a field. Will it be sort in the index, or are we only talking about sorting in memory.

I am very new to work with ElevateDB. I have many years experience working with both Delphi and C++, so if I can help to a general fix, I will be available for sure.

Do we know how long time it will take to make a decision for a patch. And if elevatedb will make this patch, when can we expect something ?! I am asking of course, so we can make a decision about what we will do, because this feature is critical for our solution.

I am very grateful for the proactive approach to solutions you share in here.

Thanks Peter

PeterBeyer

>Wow, I'm really impressed with the support of this task - thank you. Although it sounds like I'm being unemployed But I will for sure prefer if we can make a general solution.
>
>I am a little bit unsure how this will work in practice. it sounds like a switch/attribute we can set, if we want another way of sorting a field. Will it be sort in the index, or are we only talking about sorting in memory.
>
>I am very new to work with ElevateDB. I have many years experience working with both Delphi and C++, so if I can help to a general fix, I will be available for sure.
>
>Do we know how long time it will take to make a decision for a patch. And if elevatedb will make this patch, when can we expect something ?! I am asking of course, so we can make a decision about what we will do, because this feature is critical for our solution.
>
>I am very grateful for the proactive approach to solutions you share in here.

We are just users, Tim is the author and he's the only one who will make a decision about adding in another switch. You are best contacting support directly to ask. We users cannot answer you.

Roy Lambert

Roy Lambert wrote:

> We are just users, Tim is the author and he's the only one who will
> make a decision about adding in another switch. You are best
> contacting support directly to ask. We users cannot answer you.

Indeed - we can attempt to offer suggestions on how you might solve it,
but what can be done in EDB is not our decision at all. My aim was to
suggest what I'd say would be the ideal. Check with support@ on whether
this is possible in any way first. Then we can come back to alternative
mechanisms...

(It may also be worth finding out how it was done in DBISAM - what was
the solution for that?)

--

Matthew Jones

Matthew


>(It may also be worth finding out how it was done in DBISAM - what was
>the solution for that?)

They hacked the source

<<I have a task to find out, how we can add a new datatype in ElevateDB. We have done it with DBISAM>>

Roy

Roy Lambert wrote:

> They hacked the source

Sure, but what was the change made? Did they add a new collation
perhaps?

--

Matthew Jones

Matthew


You're getting lazy - from a post a while back

<<In our existing solution, I can see, that the basic funtions is made in dbisamlb.pas, in the procedure EvaluateBinary. The first line is a procedure we have made, that test if parameter Token2.DataType = ZString and Token2.SubType = SUBTYPE_FIXED. After that it test to see if Token1.DataSize < 11. If this is correct, we call a routine that transform the character data to something that can be sorted correctly in a index.

It makes sence - because the special Admind field have a charistica of size 10 and is a fixed char.>>

Roy Lambert

Roy Lambert wrote:

>  If this is correct, we call a routine that transform the character
> data to something that can be sorted correctly in a index.

That though is the key. Reminds me of the old cartoon:

http://star.psy.ohio-state.edu/coglab/Miracle.html

(I have that on a T-shirt somewhere!)

"All that needs to be done" is to apply that transform in the
application code before it is stored... Which is sort of what has been
said already. The key is whether EDB has a way to inject a transform
into the indexing system, which could either be custom or a custom
collation.

--

Matthew Jones

Okay, I understand.

It would have been nice, if we'd find a general solution that could meet our needs.

The solutions with the collation is interesting, if it were a feature that were available in elevatedb. But I guess it's not. We can be ambitious, and try to implement it our self, but it gives no meaning, if it is not elevate software that make this implementation. We will have the same problem next time, when we receive a new version of the elevatedb, as we have today.

If Elevate software are going to implement a solution, we do not know when it will be, or if it ever will be.

So right now, says my feelings, that we will move our existing implementation (that have been working well for many years), into the elevatedb framework, and take it from there.

Your support has confirmed to me, that there probably is no better option right now.


Thanks Peter

Matthew


>http://star.psy.ohio-state.edu/coglab/Miracle.html
>
>(I have that on a T-shirt somewhere!)

Wish I had - I really like the sentiment.

Roy

On 7/2/2015 9:27 AM, PeterBeyer wrote:
> If Elevate software are going to implement a solution, we do not know when it will be, or if it ever will be.
> So right now, says my feelings, that we will move our existing implementation (that have been working well for many years), into the elevatedb framework, and take it from there.
> Your support has confirmed to me, that there probably is no better option right now.

Peter,

I'd say that might be an assessment based on inadequate information.

You really should ask Elevate Software direct (thru support email).

They are very quick at responding and you'll have an answer tomorrow


Raul