detecting new files

Hello all,

I'm wondering if anyone has a more elegant solution to a pretty simple problem... does anyone have any ideas how to continuously monitor a directory for new files? The only way I can think of right now would be to poll a directory, process the file, delete it, and then sleep for a bit and check the directory again.

I mean, it works, but I can't help the feeling there's a better way of doing this... any ideas?

Thanks in advance!

OS ?

Whoops... sorry about that... used to Java and the whole 'platform independent' thing...

Looking for a Solaris solution...

In that case, maybe you want to look at the Sendmail source code - I'm not exactly sure how it works, but I assume it has some kind of algorithm that you describe for it's mail pool.

>continuously monitor a directory for new files?

New files or updated files? If new files, then here's one idea:

First count the files in the directory, sleep for a bit, then count the files in the directory again.

It depends on the platform on which you want to implement this.
On Unix and associated flavors you could use a thread that starts at system startup and then forks a child process everytime a new file is created in the directory and this child process could take appropriate action as required.

Hope this suffices.

Hey all,

Thanks for the ideas!

>>First count the files in the directory, sleep for a bit, then count the files in the directory again.<<
This won't work if one file gets deleted and another gets created, as you'll still have the same amount of files at the end.

One trouble with polling for files is that you never know if the file creator has actually finished writing to it. You could easily end up processing half a file, and confusing the hell out of the creator process.

To get round this problem (in the scripting world, as opposed to C), I've used files with specific names that denote that another file is ready for processing. For example, the data file might be called mydata.dat, and when the creating process has finished with it, it simply touches an empty file called mydata.dat.ready. This still means a polling process, but you'd only be looking for files named *.ready.

why not just create the file, wirte the data to it, and when finished, rather than create another file, just rename the existing file?

Quote:

---

Originally posted by _Elixia_
why not just create the file, wirte the data to it, and when finished, rather than create another file, just rename the existing file?

---

Yeah, that'd do. My suggestion was based on having the ability to have multiple files (different names) being created, and a .ready file denoting that the batch of files is ready, instead of a single one. I think I should'a said that before :o

Something like

Code:

---

Names.dat
Addresses.dat
PhoneNumbers.dat
AllFiles.ready

---