Hi Viktor,
My test cannot fail with the current version of TGET since :Pos()
will not allow the assignment to :nPos of a non-editable position.
Therefore, this issue is closed. HOWEVER...
I do not like to see such fragile code - The loop in :DelEnd() is
highly dependent on other methods in order to not fail (ie. the
:Pos() method). Some future change could cause this loop to go into
an endless state again.
Therefore, IMHO, protection against this should be added to this
loop. IOW, any loop should have a clear exit strategy which is NOT
dependant on code outside of the scope of that method.
Regards,
Randy.
At 10:44 AM 6/26/2008, you wrote:
Hi Randy,
Okay, that clears it up.
Before implementing your fix, I think it would be
best to find such a test case which exploits it.
This way maybe we can even find a better fix on
lower level.
Brgds,
Viktor
On 2008.06.26., at 15:15, Randy Portnoff wrote:
Hi Viktor,
I am very sorry, I just realized I am using an older version of TGET
- In the current TGET, the :Pos() method does not allow the
assignment of a non-editable position to the :nPos var - Therefore,
in the case below, when you assign o:pos := 1, it is actually
assigning 2 and thereby protecting the loop in :DelEnd().
Again, sorry... However, I still feel that the loop in DelEnd()
looks dangerous!
Regards,
Randy.
At 08:17 AM 6/26/2008, you wrote:
Hi Randy,
I've added a small test case, but I couldn't reproduce the loop:
cStr07 := Space(20)
SetPos( 14, 16 )
o := _GET_( cStr07, "cStr07", "(999) 999-9999",, )
o:display()
o:setFocus()
o:pos := 1
o:delEnd()
What am I missing?
Brgds,
Viktor
On 2008.06.25., at 19:27, Randy Portnoff wrote:
Hi Viktor,
Ok, I think I know what is happening here and is partially my fault
- However, I still feel that extra protection needs to be added to
this loop.
The picture is "(999) 999-9999"
When this endless loop occurs, I am doing this:
oGet:Pos := 1
oGet:DelEnd()
I say this is my fault since I am setting :Pos to 1, but the first
position is not an editable position.
So, what is happening is that nPos is getting set to 1 in :DelEnd()
(due to my assignment above) and then ::Pos is stopping at 2
since :Left() in :backSpaceLow() cannot get to 1 (due to the
picture) and hence the endless loop.
So, I fixed this in my Harbour build by checking for a change
to ::nPos as follows:
METHOD delEnd() CLASS Get
LOCAL nPos, nLast
IF ::hasFocus
nPos := ::nPos
::pos := ::nMaxEdit
nLast := ::nPos + 1
::deleteLow()
DO WHILE ::nPos > nPos .and. ::nPos < nLast
nLast := ::nPos
::backSpaceLow()
ENDDO
::display()
ENDIF
RETURN Self
At 12:10 PM 6/25/2008, you wrote:
Hi Randy,
Probably. If we can get a small test case
together, I will add it to rto_get.prg, and
I can try it with Clipper too.
So, var is 20 chars string, picture is "(999) 999-9999".
What keys do you press to get into the endless loop?
Brgds,
Viktor
On 2008.06.25., at 18:04, Randy Portnoff wrote:
Hi all,
I think there is a bug in TGET's delEnd() method - The loop...
DO WHILE ::nPos > nPos
::backSpaceLow()
ENDDO
...can become endless in some circumstances. While I cannot
reproduce it in a DOS app, my GUI app (that implements its own
masked edits using TGET) can using a 20-character character
variable
and a picture as "(999) 999-9999" - If I reduce the variable to 14
characters (ie. to match the picture length), it seems to work ok.
So, I'm not sure if this should be classified as a bug or not.
Regards,
Randy.
P.S. The use of :Pos and :nPos is very confusing in TGET.PRG
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
