Re: readdir ifs command

[Scott Klement <rpg400-l@xxxxxxxxxxxxxxxx>]

> However, I now need to process each file in its arrival sequence, 
> I was wondering, if anybody knows the precedence in which the readdir 
> command process each file...

I don't think there is any guaranteed order. It's whatever order the 
directory entries happen to be in the directory object... I'm not sure 
that the order is completely predictable.

> If not, I will need to read the directory once, to grab the filenames, 
> and the timestamp... and then re-read the files using that order...

I'd load the filenames & timestamps into an array, sort by the timestamp, 
then use that array to determine which filenames you process. That seems 
like it would be pretty safe.


> But then the question comes, what variable holds the best timestamp 
> value...for a stream file...

Depends on your requirements.  Personally, I'd go with the "change time" 
(the date/time the file was last changed) as my indication of whether the 
file needs to be processed.  But, I don't know if that fits in with your 
requirements or not.  You can get the "change time" field using the stat() 
API.

If you really want to go in the order in which they're created, the 
Qp0lGetAttr() API can be used to retrieve the create date/time.  This API 
is much more difficult to use than stat(), so I'd avoid it unless there's 
a really good reason to use the create date instead of the "change date".

For the projects that I've done, the "change date" made more sense, 
anyway.

Another tip:  You can unlink() a file while you've got it open (with the 
open() API).  If you do that, you can still use the file, even though it's 
gone from the directory.  The actual data portion of the file will remain 
usable until you close the file, at which point it'll be deleted.

This is useful when you want to process one file and make it completely 
unavailable to other processes on the system, so that they won't also try 
to process it.

Good luck