On Friday, May 17, 2002, at 01:06 , Scot Needy wrote:
We both agree that IF the code got into main memory -
then the 'compile' is finished - and it will just cruise along.
{ some folks have seen the 'problem' that they wrote a CGI that
forks, and found something wrong with it - removed the 'text' off
of the web server - and 'the code was still running' - so yeah,
perl beats /bin/sh all to death on that point. }
My concern here is how does it deal with the 'dynamically loadable'
'perl library' side of the game -
eg: cf p296ff with regards to Audtoloader and of course the
Autosplit - that will have generated the foo.al files...
{ cf also perldoc SelfLoader
"
The SelfLoader will read from the FOOBAR::DATA filehandle
to load in the data after `__DATA__', and load in any sub-
routine when it is called. The costs are the one-time
parsing of the data after `__DATA__', and a load delay for
the _first_ call of any autoloaded function. The benefits
(hopefully) are a speeded up compilation phase, with no
need to load functions which are never used.
"
}; # oh hell it's just an autonomous closure let the english lit
# freaks whine - it makes me feel safer....
{ man who comment email has other issues to resolve.... }
[..]
>> { is this the part where we note that DynaLoader itself
>> is 'exposed' and that Text::ParseWords as well as GetOpt::Long,
>> hence one is limiting the 'host local' coding options that
>> you might wish to have for dealing with checking this host
>> for a given problem if the NFS mount is a part of the problem? }
>
> Um not sure I understand that one. I am by every definition a beginner
> in this perl language thing :) Reading Learning Perl now to understand
> how to leverage this treed hash that Kstat has.
Well we're in the Same Boat - so let's see if we can shake one
of these 'power users' up to see if they can give it up.
I just figured out that the whole POSIX::* hierarchy is a
screaming meemee of *.al autoloader files.
GetOpt::Long - you really gotta do the PerlDoc on that one
as it will allow you really cool 'gnuIsh' tricks
myCoolPerl --file=JoeBob --Bogart=joint,bong,water
Text::ParseWords - saves on ugly lifting in doing 'expects'
like "so what did I get in that response?" sorts of questions
all of which have at least one autoloadable function -
all of which are at risk in your core monitoring system....
DynaLoad is what you need for dynamic loading - and that is
most likely gonna scram any code that calls for a FOO::BAR
module that would need dynamic loading.... because we're
all waiting on the NFS mounts to chill so that we can read
the dynamically - but not locally - mounted portion of the
code to finish the compile time events ....
> But I confess....
>
> The modules directory is local but mirrored to every system using
> Tivoli.
eg: to the best of your knowledge all the *.al files would be
disk local at that point and not coming in over NFS???
That would seem to be cool - as long as none of the XS based
code using the *.so and *.bs files - are not gonna hammer you
at run time in the NFS cross mount problem.
> My Master needs to have source for NT, Solaris, Linux and OSF. I
> already know that my destination host is of type Solaris but It would
> take a little more work to figure out rev so I just push Solaris
> Modules. Until now there hasn't been a need to split them up.
the AutoSplit is what I am worrying about - and trying to go
to an nfs based solution may be putting you at risk.
dude scope out Config - way makes life simpler on knowing what
the version of Perl Thinks is the reality - cf:
http://www.wetware.com/drieux/CS/Proj/PID/
if you want to see that in use for articulating the distinctions
in say commandline arguments to ps for linux/solaris/darwin....
ciao
drieux
---
remember boys and girls for the price of a 600M hard drive
back in 1990 you could only buy two AK-47's on the california
legal market.... { 12 in subsaharan markets }
Now you can't find either AK's or 600M hard drives -
but for the price of a 80Gig hard drive you could,
if you were in the subsaharan market pick up 2 way used AK-47's.
And to think - AK's are holding value better than hard drives.
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
