maverick-devel

[maverick-devel] [ maverick-Bugs-537111 ] MVConstantString MATCH problem

[<no...@so...>]

Bugs item #537111, was opened at 2002-03-30 09:29
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Allen Bell (apbell)
Assigned to: Nobody/Anonymous (nobody)
Summary: MVConstantString MATCH problem

Initial Comment:
Consider:

NUMB = ''
IF NUMB MATCH '1N0N' THEN
   CRT 'IT MATCHES'
END

This causes a StringIndexOutOfBounds Exception in
mvBasicString.charAt (index = 0)

It might be nice to fix this in the charAt method of 
mvBasicString because there may be siomilar ways this 
can happen, but that routine returns a char, and 
other than catching the exception from all the other 
places it might occur, I could think of no good way 
to handle it there.

In the interim, I modified mvConstantString method 
MATCH by insert a length() test at the start of the 
method to trap it. If I get time I will try and test 
other odd things to see if it can happen there.


<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>

    public mvConstantString MATCH(mvConstantString 
pattern) {

if (length() <= 0) {
   return STRING_ZERO;
}

----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

[maverick-devel] [ maverick-Bugs-537111 ] MVConstantString MATCH problem

[<no...@so...>]

Bugs item #537111, was opened at 2002-03-30 15:29
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Allen Bell (apbell)
>Assigned to: Robert Colquhoun (rjc)
Summary: MVConstantString MATCH problem

Initial Comment:
Consider:

NUMB = ''
IF NUMB MATCH '1N0N' THEN
   CRT 'IT MATCHES'
END

This causes a StringIndexOutOfBounds Exception in
mvBasicString.charAt (index = 0)

It might be nice to fix this in the charAt method of 
mvBasicString because there may be siomilar ways this 
can happen, but that routine returns a char, and 
other than catching the exception from all the other 
places it might occur, I could think of no good way 
to handle it there.

In the interim, I modified mvConstantString method 
MATCH by insert a length() test at the start of the 
method to trap it. If I get time I will try and test 
other odd things to see if it can happen there.


<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>

    public mvConstantString MATCH(mvConstantString 
pattern) {

if (length() <= 0) {
   return STRING_ZERO;
}

----------------------------------------------------------------------

>Comment By: Robert Colquhoun (rjc)
Date: 2002-03-31 06:23

Message:
Logged In: YES 
user_id=11445

I don't think the solution is quite as simple as 
automatically returning zero if the string to be tested is
of zero length.

For instance, on Unidata at least
"" MATCH "0A"
"" MATCH "0X"
"" MATCH "0B"
all return one.

There are also problems with returning an invalid character
with charAt(), if the string index is invalid you will now
need to check for the invalid character at every position
you call charAt() which in the end is the same amount of
work as checking the string boundaries before calling
charAt().  Also charAt() is a very frequently called routine
bounds checking on each call, especially when you know it is
within range, could have performance implications. 

There is a MATCH test routine in the test suite(MATCH.java
in test/org/maverickdbms/test/basic), maybe this could be
converted to basic and run on other platforms to ensure
compatibility.

----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

[maverick-devel] [ maverick-Bugs-537111 ] MVConstantString MATCH problem

[SourceForge.net <no...@so...>]

Bugs item #537111, was opened at 2002-03-30 15:29
Message generated for change (Comment added) made by rjc
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

Category: None
Group: None
Status: Open
>Resolution: Fixed
Priority: 5
Submitted By: Allen Bell (apbell)
Assigned to: Robert Colquhoun (rjc)
Summary: MVConstantString MATCH problem

Initial Comment:
Consider:

NUMB = ''
IF NUMB MATCH '1N0N' THEN
   CRT 'IT MATCHES'
END

This causes a StringIndexOutOfBounds Exception in
mvBasicString.charAt (index = 0)

It might be nice to fix this in the charAt method of 
mvBasicString because there may be siomilar ways this 
can happen, but that routine returns a char, and 
other than catching the exception from all the other 
places it might occur, I could think of no good way 
to handle it there.

In the interim, I modified mvConstantString method 
MATCH by insert a length() test at the start of the 
method to trap it. If I get time I will try and test 
other odd things to see if it can happen there.


<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>

    public mvConstantString MATCH(mvConstantString 
pattern) {

if (length() <= 0) {
   return STRING_ZERO;
}

----------------------------------------------------------------------

>Comment By: Robert Colquhoun (rjc)
Date: 2004-04-29 01:16

Message:
Logged In: YES 
user_id=11445

The rewriting of the MATCH function a month or so ago
appears to have fixed this bug.  Now ('' MATCH '1N0N') = 0
ie no match

----------------------------------------------------------------------

Comment By: Robert Colquhoun (rjc)
Date: 2002-03-31 06:23

Message:
Logged In: YES 
user_id=11445

I don't think the solution is quite as simple as 
automatically returning zero if the string to be tested is
of zero length.

For instance, on Unidata at least
"" MATCH "0A"
"" MATCH "0X"
"" MATCH "0B"
all return one.

There are also problems with returning an invalid character
with charAt(), if the string index is invalid you will now
need to check for the invalid character at every position
you call charAt() which in the end is the same amount of
work as checking the string boundaries before calling
charAt().  Also charAt() is a very frequently called routine
bounds checking on each call, especially when you know it is
within range, could have performance implications. 

There is a MATCH test routine in the test suite(MATCH.java
in test/org/maverickdbms/test/basic), maybe this could be
converted to basic and run on other platforms to ensure
compatibility.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

[maverick-devel] [ maverick-Bugs-537111 ] MVConstantString MATCH problem

[SourceForge.net <no...@so...>]

Bugs item #537111, was opened at 2002-03-30 15:29
Message generated for change (Settings changed) made by rjc
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228

Category: None
Group: None
>Status: Closed
Resolution: Fixed
Priority: 5
Submitted By: Allen Bell (apbell)
Assigned to: Robert Colquhoun (rjc)
Summary: MVConstantString MATCH problem

Initial Comment:
Consider:

NUMB = ''
IF NUMB MATCH '1N0N' THEN
   CRT 'IT MATCHES'
END

This causes a StringIndexOutOfBounds Exception in
mvBasicString.charAt (index = 0)

It might be nice to fix this in the charAt method of 
mvBasicString because there may be siomilar ways this 
can happen, but that routine returns a char, and 
other than catching the exception from all the other 
places it might occur, I could think of no good way 
to handle it there.

In the interim, I modified mvConstantString method 
MATCH by insert a length() test at the start of the 
method to trap it. If I get time I will try and test 
other odd things to see if it can happen there.


<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>

    public mvConstantString MATCH(mvConstantString 
pattern) {

if (length() <= 0) {
   return STRING_ZERO;
}

----------------------------------------------------------------------

Comment By: Robert Colquhoun (rjc)
Date: 2004-04-29 01:16

Message:
Logged In: YES 
user_id=11445

The rewriting of the MATCH function a month or so ago
appears to have fixed this bug.  Now ('' MATCH '1N0N') = 0
ie no match

----------------------------------------------------------------------

Comment By: Robert Colquhoun (rjc)
Date: 2002-03-31 06:23

Message:
Logged In: YES 
user_id=11445

I don't think the solution is quite as simple as 
automatically returning zero if the string to be tested is
of zero length.

For instance, on Unidata at least
"" MATCH "0A"
"" MATCH "0X"
"" MATCH "0B"
all return one.

There are also problems with returning an invalid character
with charAt(), if the string index is invalid you will now
need to check for the invalid character at every position
you call charAt() which in the end is the same amount of
work as checking the string boundaries before calling
charAt().  Also charAt() is a very frequently called routine
bounds checking on each call, especially when you know it is
within range, could have performance implications. 

There is a MATCH test routine in the test suite(MATCH.java
in test/org/maverickdbms/test/basic), maybe this could be
converted to basic and run on other platforms to ensure
compatibility.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102228&aid=537111&group_id=2228