Vista - Questions, questions, questions... on how to improve

Hey guys,

running my logscript has made me ponder a few things, and I have some questions
that perhaps
you scripting pros especially could help me out with.

One of them being that maaaby, I should've done the appendfile thingy after
each logfile hehe.

But you know, that's why I did the script, to learn from it. So I've thought
about it for some days,
and have some questions:

1. Is there a way to increase performance by reading blocks of data from
the logfiles, instead of
line by line ?

2. If I was to add a progress indicator, wouldn't that need to know how many
files in total it's looking
through, to be able to scale the indicator accordingly ? I'm thinking about
doing a progress indicator
for:

- Progress in relation to the current logfile
- In relation to the current folder
- In relation to the total number of files

3. I'm considering also displaying a counter for how many results it's found.
Is there a way to statically
define a variable that's shown on screen, and when the value is updated,
it's updated on screen, but
not in a new screen position ? (I know Rexx had this in OS/2)

4. Also considering displaying some metrics. How much data, how many files,
how many lines exactly,
etc. Any obvious cmdlets or methods to use here, other then getting filesizes,
counting lines in a log-
file, etc. ?

Hope you can help me out with those questions. It's funny how when you've
made a script like that, and
you're running it (especially on such amounts of data), the obvious lacks
tend to get in your face pretty
quick

But it's all good, I'm hoping to learn from it, and I'm hoping you can help
me out a bit, learning

Best Regards,
Jacob Saaby Nielsen
mailto:jacob.saaby@xxxxxx

Wow...

Noone has any input on these thoughts ? I would think that everyone could
learn from the answers, but perhaps
I was wrong.


Best Regards,
Jacob Saaby Nielsen
mailto:jacob.saaby@xxxxxx

Jacob Saaby Nielsen wrote:Well, a function versus a filter can change the performance of how
objects are read in.

If you're doing:
get-content file|some_function
or
get-content file|some_filter

A filter is acting on each object (each line) as it is retrieved by
get-content, and passing that on. A function will wait for get-content
to read the entire file.

It is a matter of testing as functions and filters are basically
structured the same way.

Will reading blocks do better? I think a filter will do a better job at
reading bigger log files, and I'll try to demo that today/over the
weekend maybe.

Marco

--
Microsoft MVP - Windows PowerShell
http://www.microsoft.com/mvp

PowerGadgets MVP
http://www.powergadgets.com/mvp

Blog:
http://marcoshaw.blogspot.com

write-progress will help you here. You need to know what to pass to
write progress beforehand though.

See 'help write-progress -examples' for some ideas.

Marco

Hey Marco,

thanks. I'll try do rewrite with a filter instead My script is STILL running
(6 days
after I started it), and I see that it's because it doesn't get a lot of
CPU time by
now.

However, there are no other processes that use the cpu on that machine. Is
there
some kind of built in "if the script has run for this long, I'll downthrottle
the cpu usage" ?


Best Regards,
Jacob Saaby Nielsen
mailto:jacob.saaby@xxxxxx

Hey Marco,

I'm just playing around with my script right now. It seems if you put -readcount
on get-content,
it DRAMATICALLY increases the performance !

I have a set of logfiles I'm running my script on for test purposes. It consists
of 91 logfiles,
taking up 87mb of space.

With -readcount 500 the script plows through them in 10 seconds flat.

Without -readcount... well, now I understand why my original script performs
like a mule on the
Nürnberg Ring. It takes 9 minutes 22 seconds !!!

That's ... an insane difference ! 562 seconds without -readcount. 10 seconds
with. I ran the script
twice, just to verify that it wasn't a one time deal. 9 minutes 17 seconds
the second time without
-readcount.

Only worry I have, is that when I set -readcount to 500, there was a difference
in the output. Without
-readcount, the list was full, with readcount, 2 names were missing in the
list. But upping -readcount
to 1000 fixed it. Any take on what might've happened there ?

Seems blockreading is a good thing after all

Best Regards,
Jacob Saaby Nielsen
mailto:jacob.saaby@xxxxxx

Jacob Saaby Nielsen wrote:Good job reading the docs... ;-) I forgot about that parameter.

As for dropping lines, I will try to do a test. Could be a bug if it
can be reproduced.

Marco

--
Microsoft MVP - Windows PowerShell
http://www.microsoft.com/mvp

PowerGadgets MVP
http://www.powergadgets.com/mvp

Blog:
http://marcoshaw.blogspot.com

Hey Marco,

yeah, I guess sometimes the docs can actually be useful

Anyway, the updated version is posted here, if anyone would like to take
a look and/or be inspired.

http://www.pipforhelvede.net/post/20...Version-2.aspx

Best Regards,
Jacob Saaby Nielsen
mailto:jacob.saaby@xxxxxx