Board index » kylix » New unofficial VisualCLX patch version 3.5 released

New unofficial VisualCLX patch version 3.5 released


2004-06-05 04:25:56 PM
kylix1
I have just released the new unofficial VisualCLX patch version 3.5 that
fixes many new bugs and also repares 3.4 issus (MDI works now).
More at www.kylix-patch.de.vu
--
Regards,
Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
 
 

Re:New unofficial VisualCLX patch version 3.5 released

Thanks!!!
Andreas what about QC 8356 ?
"Andreas Hausladen" < XXXX@XXXXX.COM >wrote:
Quote
I have just released the new unofficial VisualCLX patch version 3.5 that
fixes many new bugs and also repares 3.4 issus (MDI works now).

More at www.kylix-patch.de.vu


--
Regards,

Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
 

Re:New unofficial VisualCLX patch version 3.5 released

Alexei Cioina wrote:
Quote
Andreas what about QC 8356 ?
Looks very hard to find. And I do only change VisualCLX code not RTL or DB.
--
Regards,
Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
 

{smallsort}

Re:New unofficial VisualCLX patch version 3.5 released

Alexei Cioina schrieb:
Quote
Andreas what about QC 8356 ?

Have you tried to read or set the ExceptionMask after connection? (not
tested).
 

Re:New unofficial VisualCLX patch version 3.5 released

Andreas Hausladen wrote:
Quote
I have just released the new unofficial VisualCLX patch version 3.5 that
fixes many new bugs and also repares 3.4 issus (MDI works now).
nope, MDI doesn't work here. K3 , fc1.
It opens normal window..not MDIChild inside MDIForm.
 

Re:New unofficial VisualCLX patch version 3.5 released

zeljko wrote:
Quote
Andreas Hausladen wrote:

>I have just released the new unofficial VisualCLX patch version 3.5 that
>fixes many new bugs and also repares 3.4 issus (MDI works now).

nope, MDI doesn't work here. K3 , fc1.
It opens normal window..not MDIChild inside MDIForm.
I have a small test MDI application that I use to test whether a particular
bug has been fixed by the patches. That application works fine with the
3.5 patches using k3. I have ran the app under both Libranet 8.1 and
Mandrake 9.2 and it always creates a proper MDIChild forms.
Despite the warning with the 3.4 patches that MDI was broken this
application actually worked under the 3.4 patches, however, the bug was
still present.
HTH
Gordon Hicks
 

Re:New unofficial VisualCLX patch version 3.5 released

theo < XXXX@XXXXX.COM >wrote:
Quote
Alexei Cioina schrieb:

>Andreas what about QC 8356 ?
>

Have you tried to read or set the ExceptionMask after connection? (not
tested).
...
var c:Currency;
d:Double;
...
SQLConnection1.Connected:=true;
Set8087CW($133f);// This workaround looks very stupid for me.
d:=0.0;
c:=1.0/d;
ShowMessage(CurrToStr(c));
// e.g. c=-922337203685477.5808 is very stupid for me.
 

Re:New unofficial VisualCLX patch version 3.5 released

Alexei Cioina wrote:
Quote
SQLConnection1.Connected:=true;
Set8087CW($133f);// This workaround looks very stupid for me.
Could it be that the DB driver DLL changes the coprocessor word?
--
Regards,
Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
 

Re:New unofficial VisualCLX patch version 3.5 released

"Andreas Hausladen" < XXXX@XXXXX.COM >wrote:
Quote
Alexei Cioina wrote:

>SQLConnection1.Connected:=true;
>Set8087CW($133f);// This workaround looks very stupid for me.

Could it be that the DB driver DLL changes the coprocessor word?


--
Regards,

Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
No!
 

Re:New unofficial VisualCLX patch version 3.5 released

Sorry. I had uploaded the patch version 3.4 as 3.5 . This is now corrected
please download the patch again.
--
Regards,
Andreas Hausladen
(www.kylix-patch.de.vu - unofficial Kylix 3 patches)
 

Re:New unofficial VisualCLX patch version 3.5 released

Alexei Cioina wrote:
Quote
theo < XXXX@XXXXX.COM >wrote:

>Alexei Cioina schrieb:
>
>
>>Andreas what about QC 8356 ?
>>
>
>Have you tried to read or set the ExceptionMask after connection? (not
>tested).


...
var c:Currency;
d:Double;
...

SQLConnection1.Connected:=true;
Set8087CW($133f);// This workaround looks very stupid for me.
That should be "Set8087CW($1332);" to allow FPU exceptions to happen.
Quote
d:=0.0;
c:=1.0/d;
ShowMessage(CurrToStr(c));
// e.g. c=-922337203685477.5808 is very stupid for me.

That's not stupid at all. Currency is handled as a 64-bit integer
internally. From the intel Processor Manual, Instruction Set Reference,
concerning fist/fistP (FPU store integer instruction):
------------------------------------------------------------------------
If the converted value is too large for the destination format, or if
the source operand is an ? SNaN, QNAN, or is in an unsupported format,
an invalid-arithmetic-operand condition is signaled. If the
invalid-operation exception is not masked, an invalid-arithmetic-operand
exception (#IA) is generated and no value is stored in the destination
operand. If the invalid-operation exception is masked, the integer
indefinite value is stored in memory.
------------------------------------------------------------------------
The latter is exactly what happens. integer indefinite values are MSBit
= 1, all other bits set to 0; in reference to the above code:
PInt64(@c)^ := $8000000000000000; // 64-bit integer indefinite
ShowMessage(CurrToStr(c)); // Result: -922337203685477.5808
Robert
 

Re:New unofficial VisualCLX patch version 3.5 released

Alexei Cioina wrote:
Quote
"Andreas Hausladen" < XXXX@XXXXX.COM >wrote:
>
>Could it be that the DB driver DLL changes the coprocessor word?
>
...
No!
How do you know? Andreas' explanation seems natural to me.
Robert
 

Re:New unofficial VisualCLX patch version 3.5 released

Robert Rossmair < XXXX@XXXXX.COM >wrote:
Quote
Alexei Cioina wrote:

>theo < XXXX@XXXXX.COM >wrote:
>
>>Alexei Cioina schrieb:
>>
>>
>>>Andreas what about QC 8356 ?
>>>
>>
>>Have you tried to read or set the ExceptionMask after connection? (not
>>tested).
>
>
>...
>var c:Currency;
>d:Double;
>...
>
>SQLConnection1.Connected:=true;
>Set8087CW($133f);// This workaround looks very stupid for me.

That should be "Set8087CW($1332);" to allow FPU exceptions to happen.

>d:=0.0;
>c:=1.0/d;
>ShowMessage(CurrToStr(c));
>// e.g. c=-922337203685477.5808 is very stupid for me.
>

That's not stupid at all. Currency is handled as a 64-bit integer
internally. From the intel Processor Manual, Instruction Set Reference,
concerning fist/fistP (FPU store integer instruction):
------------------------------------------------------------------------
If the converted value is too large for the destination format, or if
the source operand is an ? SNaN, QNAN, or is in an unsupported format,
an invalid-arithmetic-operand condition is signaled. If the
invalid-operation exception is not masked, an invalid-arithmetic-operand
exception (#IA) is generated and no value is stored in the destination
operand. If the invalid-operation exception is masked, the integer
indefinite value is stored in memory.
------------------------------------------------------------------------
The latter is exactly what happens. integer indefinite values are MSBit
= 1, all other bits set to 0; in reference to the above code:

PInt64(@c)^ := $8000000000000000; // 64-bit integer indefinite
ShowMessage(CurrToStr(c)); // Result: -922337203685477.5808

Robert

I understand that variable c looks like integer indefinite values but algebraic rules for Nan do not work in this case. E.g for Double we have Nan+1=Nan, Nan-1 =Nan etc. , For currency
Nan-1=Nan (-922337203685477.5808), Nan+1= -922337203685476.5808<>-922337203685477.5808. This means that -922337203685477.5808 is a good value for operating with it!
Also do you like this?
Set8087CW($133f);
VarComplexCreate(Nan,Nan); // ->0+0i
 

Re:New unofficial VisualCLX patch version 3.5 released

Robert Rossmair < XXXX@XXXXX.COM >wrote:
Quote
Alexei Cioina wrote:

>"Andreas Hausladen" < XXXX@XXXXX.COM >wrote:
>>
>>Could it be that the DB driver DLL changes the coprocessor word?
>>
...
>No!

How do you know? Andreas' explanation seems natural to me.

Robert
I tried libbdwxpUIBfire15.so driver !
 

Re:New unofficial VisualCLX patch version 3.5 released

"Alexei Cioina" < XXXX@XXXXX.COM >wrote:
Quote

Robert Rossmair < XXXX@XXXXX.COM >wrote:
>Alexei Cioina wrote:
>
>>"Andreas Hausladen" < XXXX@XXXXX.COM >wrote:
>>>
>>>Could it be that the DB driver DLL changes the coprocessor word?
>>>
>...
>>No!
>
>How do you know? Andreas' explanation seems natural to me.
>
>Robert

I tried libbdwxpUIBfire15.so driver !
libdbexpUIBfire15.so
sorry