Converting Characters with PB 12.6 - CharA not working correctly?

[Former Member]

Hi,

i am just testing a migrated Powerbuilder application from PB 9.5 to 12.6. here we have some functionality to read data from a Blob and show it later on with a datawindow.

In this context i discover currently a strange behavior.

e.g.

// Get record type (1 Byte)

lstr_record.record_type = CharA(Blobmid(lb_block, 3))

This code is working in Powerbuilder 9.5 (only Char instead of CharA) and Powerbuilder 12.5 (which i tested some weeks ago), but in Powerbuilder 12.6 i get some strange chinese character.

I was trying to get deeper into the issue and have converted the Blob into Byte Array and i looked to the 3rd byte in above case and i got 67 = 43hex, which would be the expected character 'C'.

Is this a bug in Powerbuilder 12.6? Or what do i missing?

[Former Member]

Hi Henning,

the problem seems to be the implicit conversion from a Blob with length 1 to an integer. Try to add a lblb_00 which is declared as "blob{1} lblb_00".

... = charA( BlobMid( lb_Block, 3, 1 ) + lblb_00 )

charA() takes a string or an integer, in this case obviously an integer and an integer is 2 bytes long.

HTH

Bernhard

[Former Member]

Hi Bernhard,

i did not tried yet you suggestion, but i found now another way, which works fine.

Solution looks like this:

// Get record type (1 Byte)

lstr_record.record_type =  String(Blobmid(lb_block, 3,1), EncodingANSI!)

This will solve my issue, but anyway strange to see, that code, which is working in Powerbuilder 9.5 and Powerbuilder 12.5,  is not working anymore in Powerbuilder 12.6.

Regards

Henning

[Former Member]

Hi Henning;

  FWIW: If  this exact code worked in PB 12.5.x then I suspect that its a bug in PB 12.6 as the code base between the two PB versions should be the same - AFAIK.

Just for interest sake though, can you try ...

lstr_record.record_type = CharA ( Blobmid ( lb_block, 3, 1 ) )

Regards .. Chris

[Former Member]

Hi Chris,

That i tried too, same result

This code does not work:

// Get record type (1 Byte)

lstr_record.record_type = CharA(Integer(Blobmid(lb_block, 3,1)))

This code gives the expected result !!!

Byte lb_byte[]

lb_byte = getByteArray(Blobmid(lb_block, 3, 1))

lstr_record.record_type = CharA(Integer(lb_byte[1]))

It looks, really like a bug. For me it seems to be, that CharA reads 2 bytes instead of only one

Henning

[Former Member]

I agree... I think that either BlobMid or CharA are reading two bytes instead of one. This certainly looks like a case should be opened with SAP technical support for this.

I'm glad though that you have a workaround though using the GetByte/GetByteArray () methods!