Subject | Re: Trying to run TrackStudio :-) |
---|---|
Author | the_a_rioch |
Post date | 2012-06-29T08:19:51Z |
> The issue with Java is, that every byte sequence has to be converted toIn Delphi you has also option to have char array/pointer.
> a Unicode characters. That is Java. Period. Or access them as byte
> array. Another period. :)
And after Delphi switched to Unicode - it is a messs... :-)
> With your Delphi background you can think of having only wide stringsNo-no-no, OEM2ANSI is different beast - it is transcoding SBCSes, AnsiString in Delphi terms. But i understand what u meant. Since Windows is runing in two different russian encodings at same time - how can anyone from Russia forget those issues :-D
> and doing oem-to-ansi conversion each time you get some raw data from
> disk or network.
> In order to convert a byte steam into a string youAnd those assumptions should match at both sides of pipe.
> have to make assumptions re in which encoding data is coming in.
I'd say they should be enforced to match.
> TheConenction string ? Is it the thing we pass to FBClient.DLL, so it is exposed in data access components ? Or, respectively, connection string is JDBC URL.
> encoding parameter in the connection string tells the server, in which
> encoding we want to receive data,
I think that there is no connection string in the internal protocol used between server core and fbclient.dll/jaybird
There are probably some structs or sequence of API calls, but not that "connection string" ?
Just out of curiosity, it is not realyl important HOW is that property transferred.
But surely server SHOULD be somehow informed which connection charset attached client assumes to have.
> When you specify NONE encoding, then you explicitly tellStop here. That is not true!!!
The situation is not when i "explicitly specify NONE".
The situation is when i, having lack of knowledge, did not specify anything.
So the correct wording would be:
"When user does not specify NONE encoding, then he implicitly tell
server to perform no translation and hope that client will get it
right."
Yeah ? i implicitly tell *database* server to hope. Implicitly. To hope. Kill me baby.
> right. It is a powerful mechanism to solve some problems inExactly. And as such this mechanism should NEVER be engaged by accident. Only purposedly.
> heterogenous environments, but when you use it wrong, you can make
> database unusable.
And curently, if connection is not specified, you treat it as specified NONE.
And then FB chooses to play-by-rulebook.
And then JB chooses to play-by-guessing.
Hilarity ensues. :-/