Re^2: RFC: Abusing "virtual" memory

[sundialsvc4]

Good sir or madam, I wouldst delicately and most-diplomatically point out that it is, in fact, not so that your intuitive suppositions are correct ... but that, in fact, it is precisely my key-point that, (counter-intuitively though they may-or-may-not be), they are not.

For at least a generation, our predecessors made-do without computers of any kind... they used only punched cards, and with those primitive tools they accomplished miracles.

For another full generation beyond that,computers existed, yes, but copious memory-sizes did not. “A million bytes of memory?! Gosh!”

It is, indeed, precisely(!) my point that our grandfathers with their punched-card machines had, and still have, a valuable lesson to teach us.

[jbert]

Sorry, I have no idea if your two replies to me in this thread are as a result of my having offended you or not. If you're offended I'm sorry, that wasn't my intent.

I'm probably being dense, but as far as I could see, neither of your responses answered the question:

"If you need to process the keys in sorted order - and the volume of data of the keys is significant - why are you storing them in a data structure which does not keep them in a sorted order?"

Your oblique references to performance perhaps suggest that you think the cost of updating the index on insertion will be too high to meet your throughput needs.

Is this right?

If there's some other meaning, I'm afraid I've missed it. Perhaps you could rephrase it?

Update: is this all a joke, btw? 44Mb of RAM isn't a significant amount out of the buffer cache of "a very expansive computer, plenty of memory, fast CPU". Or is this all on a historical machine? Or is your point something about human versus computer computations?