Re: st: timestamps, GNU/make-style Stata program

[James Muller <james.muller@internode.on.net>]

Okay, I'll do this properly then.

My feeling is that Dan is right, that it should be compiled. Returning a 
timestamp is a sufficiently fundamental task that it should be quick as 
possible. It is surprising that nothing like this exists in the standard 
Stata distribution - perhaps Statsoft could pick it up?

Further, it would be nice to implement a fuller/alternate range of OS 
functions that perform more of the Unix/Linux-style tasks. For example, 
a program to return the file permissions, or a program to recursively 
delete a directory structure, or an "ls" that returns the directory 
contents in r(). Most of these sorts of things are not terribly hard to 
implement and they seem to open up Stata quite a bit.

And then comes the Make. I won't do anything about all this for a few 
days, but will look into writing a proper make utility for Stata. It 
would be really neat to be able to stick to standard GNU/Make syntax, 
but that's a major effort. It may even be possible to modify the 
GNU/Make code to interface with Stata in its current form, though this 
might create a problem of OS-dependence; will see.

Yes, it is true, GNU/make, etc. already exist, are fast and are 
well-evolved, but they work within the host OS (the one running Stata). 
My view is that Stata itself is an operating system, so it would be nice 
to have similar functionality completely within it (_not_ using the 
system command).

Give me comments/suggestions/thoughts.

After I've delved into this a bit more I'll write a follow-up. I'll also 
be asking for people to compile binaries on different OS's, and if I do 
the Make program, will need testers on as many supported Stata host OS's 
as possible - for the future.

Cheers

James

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/