Primarily though, this BugTracker is used to track the features/projects/ideas for my Xtools project.  I figured that while a diary is great, tracking future bugs and major progress in the project would be a little more automated if I utilized a BugTracker.  MantisBT is fairly full featured.  I have found a few bugs, and submitted them to the developer’s bug tracker.

Tasks completed:

• Finished the basic DNB Parser
• Finished the basic DNB Display page
  • Added simple filter and sort options
• Finished the basic Batch Upload page

I recently tallied the lines of code I’ve written in ASPECT scripting and I’ve reached over 14,000 lines of code including comments, just under half of that is actual code, which means that I’m doing a fairly good job at commenting my code to make it more readable.  A significant portion of the comments include the header of each module, which includes the change log to each module.

Back then Wikipedia didn’t have an article on bitwise operations and I did not have the access to mathematic books.  I was doing some coding this last weekend and found an irregularity which I have still not successfully reproduced or explained, and while doing research on the matter, I came across the bitwise operation article at Wikipedia and discovered why ASPECT does not work as intended:

The reason for this is that C uses a logical shift whereas ASPECT uses an arithmetic shift.  This plays havoc with RFC 1321 and as such I had to do some mathematical tinkering to get an arithmetic shift to behave like a logical shift, which allows for the creation of a rotate no carry shift.

This is the code that comes from RFC 1321

#define ROTATE_LEFT(x, n) (((x) << (n)) | ((x) >> (32-(n))))

This is the resultant code with the arithmetic to logical shift conversion code:

func ROTATE_LEFT : integer
 param integer x, n
 integer z
 z = (x >> (32-n)) & ~(0×80000000 >> (31-n))
 return ( (x << n) | z )
endfunc

In order to reduce compiler and stack overhead, this could probably be shorted to a #define that looks like this (I haven’t tested this):

#define ROTATE_LEFT(x,n)  ( (x << n) | (x >> (32-n)) & ~(0×80000000 >> (31-n)) )

In summary you replace

((x) >> 32-(n))

with

 (x >> (32-n)) & ~(0×80000000 >> (31-n))

This causes the arithmetic shift to operate like a logical shift if the sign bit is set to 1 (which is the only time an arithmetic and logical shift operate differently).  I posted a bit of source to my script page which demonstrates this “bug” and the fix.

NOTE: A circular shift (or rotate no carry shift) is not inherent in either C source or ASPECT, hence the creation of the ROTATE_LEFT function in ASPECT source (or the compiler #define in C source)

Merry Christmas

The best part about it is that it’s extensible.  Meaning that back in 2003, I did some coding using the C++ language to write a DLL API to further extend the functionality of Procomm.  Specifically, I added the ability for Procomm to access POP3 email via a API call to a DLL I wrote.  The idea was to retrieve instructions to perform certain tasks via email, perform them in an automated fashion, and then return to log to the person who requested the task via email.  Heck, if you wanted, you could write rlogin DLLs and have Procomm emulate rlogin via the DLL functionality.

Sadly, I lost my dedication to the projects I was working on because I didn’t have access to a Meridian-1 system full time to do testing with.  Later I did get a Meridian-1 second-hand, but I never recovered the urge to continue building the tools (other enjoyable tasks to pursue.)  Now, I’m finding the drive again.

1. Recognize Canadian area codes (to permit Canadian users to add all US and Canadian NPAs but exclude all other NPAs).
2. Added protection against NPA.TXT corruption: Validate NPA value from NPA.TXT before sending value to SL1 interface.  (Value must be between 200 and 999).
3. Updated NPA.TXT to include all area codes created since last update.  Still excluding 800, 866, 877 and 900 NPAs to permit these to be manually configured as SPNs.
4. Updated X11BARS documentation.

Thank you Andrew Chapelle for the bug report.