Hi,

when trying to us a TSdfDataset connected to a TDatasource and a TDBGrid, only some unicode symbols are shown (those little squares holding four tiny numbers) for string data.

The data files are encoded iso-8859 but I have found no switch for setting the encoding of TDataSource or TSdfDataset. When loading a TFileStream is used, but I did not find a property or method for changing the encoding on the fly in it, either.

How can I fix it?

TIA,
Marc

In my limited experience depends critically on the version etc. of the OS you're running.

I don't understand. I'm using FreeBSD, but the program will run on Windows.

I'm seeking a hook for attaching sth. like "new := ANSI2Unicode(old)", or a switch for changing the expected encoding in one of those components involved. Man .csv importers do ask for the encoding, Openoffice does.

The alternative would be to parse the file and convert each string piece by piece, showing in a simple TStringGrid - I don't like that idea. ;)

Digging some TSream*-Classes in Unit Classes I think one way to go would be to make a TString-descendant of some type, maybe TCustomMemoryStream or TStringStream and add a method "ReadUTF8()" or similar. Or directly override the Read()-Method for returning converted Strings.

Still hoping for a more simple solution, any ideas or hints?

TIA,
Marc

t

be to make a TString-descendant of some type, maybe TCustomMemoryStream

TStream-

None. It is an old problem, see:

http://bugs.freepascal.org/view.php?id=16587

This can be easily solved when we start the Unicode RTL and classes.

Michael.

OK, at least I'm not writing code already done.

Thanks,
Marc

Got to ask some more questions ...

How would that easy solution look like?

From digging the code and thinking about it it is clear that my

implementation has to be inside TSdfDataset or a descendant. It cannot be done using a special file stream because of the byte count change with UTF8.

How can the current system encoding be retrieved in a platfrom neutral way?

Would it be OK to only stick any string into Ansi2utf8() and hand over the result in this case? Here I know the files encoding and it is fixed to iso8859-1, no change ever.

But a general solution would have a property storing the files encoding and a list of known encodings, do fpc/lazarus have knowledge of encodings somewhere?

Forget about it. Too small budget, too much work. Stopped.

Marc