Hello andrew!

08 Jun 18 20:36, andrew clarke wrote to Gerrit Kuehn:

 GK>> That's what I was referring to. Of course there are ways around that, 
 GK>> so I wrote "not easily", i.e., not the usual way.

 ac> Well, not the DOS way. :-)

Not the shell way, either.

 ac>>> find . -iname '*.msg' -delete

 ac> Actually this will delete *.msg in subdirectories too. And also 
 ac> subdirectories with names matching that pattern.

Only if they're empty.

 ac> Use this instead:
 ac> find . -depth 1 -type f -iname '*.msg' -delete
 ac> Omit -delete if you just want a list of files that it will delete.

However, you'll still run into trouble when having many files: find might
just eat up all your ram, so you're back to writing a shell loop processing
files.

 GK>> Might take for ages, too. find is slow with that. I've had 
 GK>> directories with many files, and even doing "ls -l"
is getting slow 
 GK>> pretty soon. Try it with 50k of files, and you'll see what I mean.

 ac> This hasn't been my experience on modern PCs. Especially not with 
 ac> journaling filesystems on an SSDs.

Did you actually try? In my experience, especially "modern" file
systems tend to cause all kinds of trouble on operations like this.

 ac> 50,000+ messages in an echo is an edge case, though. Most echos 
 ac> rarely see that much traffic these days andtThere is not usually a 
 ac> need for anyone to keep that many messages anyway.

I have two or three echoes with 50k+ messages (one even at 120k+), and I
know quite some nodes who chose to simply keep everything decades ago (when
harddisk growths started outperforming message sizes).


Regards,
Gerrit

...  5:28PM  up 152 days, 19:21, 9 users, load averages: 0.15, 0.14, 0.09

--- Msged/BSD 6.1.2