Detecting directory changes

[Dave Jones]

In message <3E72073E.1060700 at qsl.net>,
  "John E. Malmberg" <wb8tyw at qsl.net> writes:
>The file system is scheduled for a major overhaul.  I can not give 
>details, but any program that makes any assumptions about undocumented 
>internal structures such as names of locks may break.

Yes, such a program may break in the future when someone eventually tries to
run it under some future version of VMS.  In the mean time, there are problems
to be solved now where using undisclosed (they hopefully were documented
somewhere by the developers) interfaces is the only practical solution.
Such a program should isolate and abstract the fragile pieces as much
possible.


  You do want to be 
able to support disks that are larger than a Terabyte, or files that are 
that big don't you?

>Parts of the retired POSIX product were used as a basis of the DII COE 
>effort, and the DII COE routines that are used for UNIX portability are 
>being merged into the base OpenVMS product.  As this is happening, the 
>POSIX modules may be removed or stop working.  Or they may continue to 
>be used.  This is being decided by the teams involved.


We just recently went through an issue on comp.os.vms with the CSWING 
freeware application that was and still is making invalid assumptions 
about how the file system operates or will operate in the future.  Some 
early programmer apparently got this mis-information from reading the 
source listings, or from trial and error.



Now I have put in requests for features to the file system, like a 64 
bit ino_t type, storing at file close the apparent UNIX size of a record 
oriented file for stat to return, storing the UTC time in addition to 
local time so that stat will run faster.  I have also requested support 
for alternate filenames, to help support ODS-2/ODS-5/ODS-? and 8.3 file 
lookups for applications like Pathworks Advanced Server and SAMBA.


But when it comes time to look at funding and time allocation.  Customer 
requested enhancements get priority over internal requests when all 
other things are equal.

The use of the internal POSIX routine may work in the short term, but it 
should not be considered a permanent solution.

-John
wb8tyw at qsl.network
Personal Opinion Only

PLEASE READ THIS IMPORTANT ETIQUETTE MESSAGE BEFORE POSTING:

http://www.catb.org/~esr/faqs/smart-questions.html