Bhaskar wrote:
>Code specific to each MUMPS implementation should belong in the kernel,
>isolated from the applications.
>
>GT.M doesn't provide a function to stat a file.  The two approaches
>would be to use ZSYstem to call a program like ls, stat or file, or to
>call a standard C library function to get the answer.

This reminds me of what I think might be THE most elegant way to extend MUMPS 
to a more
complete computing paradigm. Allow direct mapping of a global to a directory of 
files such
that from the MUMPS side, a file could be treated simply as a global variable, 
and from
the OS side, MUMPS data (some, not necessarily all) could be made directly 
accessible to
other applications.

When I was (prior to GT.M/Linux) actively seeking a way forward for MUMPS 
development that
could avoid lock-in to proprietary databases and trying to find a way to have 
the benefits
of MUMPS globals in other programming languages, it occurred to me that the 
closest thing
to MUMPS globals outside of MUMPS is the file system provided by each OS.

Both provide an ordered hierarchical collection of persistent shared named 
values. MUMPS
globals are generally optimized more for speed of access to a larger tree of 
data with
generally smaller data values, but the similarities seem amazingly close to me.

The primary differences are that
  1) files are essentially unlimited in size where MUMPS data values
     generally have been constrained to a size of hundreds or perhaps
     thousands of bytes. GT.M now allows local values up to one megabyte
     but limits global values to blocks of 1KB up to 32KB (last I checked).
  2) Every file generally has individual access permissions but an individual
     global reference generally does not. Some MUMPS implementations allow
     individual globals to map to separate files while others allow ranges of
     subscripts within a given global to be mapped to different files.

Providing an optional globals model that simply maps a global to a directory of 
a file
system would make the exchange of data with other applications simpler and more 
uniform, A
MUMPS data file that could be mounted as a file system for some purposes could 
provide a
potentially faster file system.

---------------------------------------
Jim Self
Systems Architect, Lead Developer
VMTH Computer Services, UC Davis
(http://www.vmth.ucdavis.edu/us/jaself)


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
_______________________________________________
Hardhats-members mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/hardhats-members

Reply via email to