Uri Guttman wrote: >>>>>> "CO" == Chas Owens <chas.ow...@gmail.com> writes: > > CO> On Fri, Aug 28, 2009 at 16:32, Uri Guttman<u...@stemsystems.com> wrote: > CO> snip > >> now, as randal also asked, why do you want this? you are just learning > >> perl and i am sure this is way above your head in many dimensions. and i > >> am positive this is an XY problem where you want X and think Y is the > >> solution when there is usually a much better way to do X. what is your > >> real GOAL here and i don't mean any technical stuff. why do you want > >> this crazy thing? > CO> snip > > CO> I believe, based on reading other emails, that he thought that cyclic > CO> data structures were leaked by the Perl program and continued to take > CO> up memory, which would mean if you could get the address of the memory > CO> you could create a set of Perl objects that would survive the ending > CO> of a Perl program and the starting of a new one. Of course, this is > CO> just a misguided attempt to recreate shared memory via a bug that > CO> doesn't actually exist (Perl cleans up cyclic dependencies at program > CO> end, you only ever "leak" memory while you are running). > > there is a deeper misunderstanding which i mentioned before. you can't > save memory addresses to disk as you can't make that data reload to the > same memory location later. PL/I had a solution for this by allowing > you to create a chunk of ram for allocation and all addresses inside it > were actually offsets from the beginning of that chunk. this allowed > that whole chunk and its complex data tree (including cycles) to be > stored to disk and later reloaded in to ram and usable. but i have never > heard of any lang doing that in years. it is easier to dump stuff in a > ram neutral format with storable, dumper, yaml, etc.
Guys, The idea started off as a way to store data in memory for a very short period of time between invocations of a CGI application. I then got thinking about how to do the same thing, without CGI. For instance, I have a Perl program that generates a hash of numbers. This program needs to exit completely before another program loads about a second later, and retrieves the data. In this case, the first program has no way of passing the hash to the second as a parameter. I was naive enough to think perhaps I could do something like this: my %hash = ( this => 'that', these => 'those' ); my $hash_ref = \%hash; # put the ref's string data in another scalar my $manual_ref = "HASH(0x8100af0)"; I knew that there were ways to temporarily store the data to disk and then re-fetch it, but not that the data could avoid the disk while still being persistent. Thanks to those for the suggestions. All in all, when I get an idea, I generally like to take the path of resistance, so I can learn for myself why something does/doesn't work. For the most part, I don't like to re-invent the wheel, unless re-inventing the wheel will teach me something important (such as being told not to do it :) or give me better understanding as to why not to do it, instead of just being told not to. It's kind of like the time I was a kid, and I was told not to touch the cow fence. By touching it, I quickly learnt by my own experience that if I take a leak at the side of a road, ensure there isn't a steel fence nearby... Steve ps. Thanks for all of the feedback. It's really help me to understand why not to do it, why it doesn't work, and alternatives to what I wanted to do. Cheers!
smime.p7s
Description: S/MIME Cryptographic Signature
