mdaems,
Your analysis is correct, except for the
I agree with that, but below the text I say that chosen field of type Varchar 255 ... what to do if the problem is exactly like that.
I think that the "temporary and crooked" decision of the forcible removal of a patch from the source (and try to do in my project). Here is the way the question: What is justified type selection TByteArray?
You can also use a dynamic array....oh ... My apologies, it's Pascal!
![Smile :)](./images/smilies/icon_smile.gif)
I write in C + +, so it did not immediately thought ...
Well, in any case, it is unclear why the use of the limited size of 32 kb array, because even with "segmented" model of memory segment size - 64 kb. And given the fact that now uses the "flat"model (indicated by as much as you need) a similar approach to the economy - even beyond my comprehension.
By the way, and how to return the data interface library database itself? After all, none of them are not written in Delphi. All of them (well, at least that I know of) written in C. And there is hardly referred to the use of such TByteArray. In any case, EMS Postgre Manager, written using ZEOS 6.x performs these queries correctly.
What does this mean that if you remove the code from the error response, then ZEOS will work properly? But in this case where in the data will be stored? (Somewhere in a pile of garbage?:))
I see what I can offer, but can not guarantee the outcome, since the main effort still at work.