Re: Inner loop compiler
Rforno,
> Perhaps you should emphasize the "may" in "may be impossible".=20
> It also "may be possible". I didn't really try to discourage you.
I understood that. :)
> What I was thinking is that your proposal means too much effort for
> only an addition to Euphoria. It seems that what you propose should
> be, in its own right, a full new compiler.
That would be nice, but then it would need tons of code for I/O, =
interfacing with OS and memory managers, etc. And I would have to worry =
about things like garbage collecting if I wanted that, to mention just =
one thing. It would amount to years of hard work!
> As an anecdote, I recall that some years ago I programmed a sort=20
> (in Commodore Basic) that I thought was performing an impossible
> task: merge in place. I was convinced it did it, and in fact it
> sorted data, but its rate of growth for the timing was steeper than
> O(n log (n)). I sent it to Donald Knuth, who never answered. About
> a month ago, I revised the code, and found that in fact the task
> was impossible: it was sorting inside the sort, instead of merging!
Forgive my utter obtuseness, I didn't quite grok that. What does it mean =
to "merge in place"? And what do you mean with "sorting inside the =
sort"?=20
Yours Sincerely,
Barbarella
|
Not Categorized, Please Help
|
|
