Migrating to Harbour

[Enrico Maria Giordano]

Elvira,

Hello,

So I beleive that the better solution will be to post those comments in the Harbour Developer forum.

Thanks.

Right. Antonio?

EMG

[elvira]

Hello,

Here is the list:
https://groups.google.com/forum/#!forum/harbour-devel

I guess the more messages more atention will bring on the topic and bug.

[Antonio Linares]

Enrico,

I appreciate if you post the msg yourself.

I already did it, so another one reporting it will benefit all of us

[Enrico Maria Giordano]

Antonio,

Enrico,

I appreciate if you post the msg yourself.

I already did it, so another one reporting it will benefit all of us

I prefer to not have to do with Harbour developers, sorry.

EMG

[Antonio Linares]

I can read your mind

[Enrico Maria Giordano]

Antonio,

I can read your mind

EMG

[James Bott]

Enrico,

Yes, just like CTOD( "" ).

Yes, but it is a workaround for the lack of being able to use nil in DBFs. 0, .f., and "" are all data, not the lack of data, but we use them that way because we have no choice. 0 and .f. are real data but since "" is an empty string it pretty much indicates no data.

This is from xHarbour hbwinole.c:

Code: Select all

     case HB_IT_DATE:
         if( pItem->item.asDate.value == 0 )
         {
            PHB_V_VT( pVariant ) = VT_NULL;
         }

Excellent, this is the evidence we need.

I searched Github for hbwinole.c and it doesn't seem to exist so I couldn't check it. There is however a hbwinole.h so I guess the C file is named something else.

As you have been saying, the only solution seems to be to convince the Harbour developers to fix this issue.

[Antonio Linares]

Done

https://groups.google.com/forum/#!topic ... rZFS5ePSnc

[Enrico Maria Giordano]

Thank you, Antonio!

EMG

[James Bott]

I also posted to Antonio's message (link shown in above message). I included Enrico's example code and the error generated.

Hope this helps.

[Antonio Linares]

Przemek already answered and it seems as it may be included in next Harbour build.

Anyhow I asked him to confirm it

James, many thanks!

[James Bott]

Przemek said:

In short words it converts Harbour empty dates to NIL so when
we add it then it will be impossible to exchange empty dates
between two Harbour applications using OLE as connection
interface.

It sounds to me like he is saying no because it will break app-to-app data exchange using OLE. Perhaps we can suggest a flag as a workaround. Something like lUseNilDates.

Note: Apparently my post to the Harbour Developer's forum is not showing up because it has to be "approved" first. I guess I am on a "watch list."

[Enrico Maria Giordano]

As usual. Great attention to theory and no care for practical problems. This is one of the reasons why I dropped Harbour...

EMG

[Antonio Linares]

James,

I just saw your msg in the list, its already there.

You may have avoid the use of #include "FiveWin.ch" as it is not needed at all for this example and they may not accept it as a reproduceable example

Maybe you could reply to yourself in the list explaining that no FWH is required to test it

[Antonio Linares]

James,

Anyhow, Przemek knows that FWH is not related to this issue