|No such thing as a small change|
Perl 5 Optimizing Compiler, Part 6: A Love Letter To Dave Mitchell, chromatic, and BrowserUKby Will_the_Chill (Monk)
|on Sep 01, 2012 at 05:30 UTC||Need Help??|
What a discussion! Thank you very much to everyone who has given their input, and all the lurkers out there, and all the Perl users (direct & indirect) everywhere - it is for you that Perl exists and that (hopefully) someday Perl 5 will be even more awesome than it is now. Also, much respect to Nick Clark, Reini Urban, Ingy, and the others who are actively mentoring me on this quest.
The aforementioned notwithstanding, this message is my love letter to the 3 most important personalities in our ongoing discussions here at Perl Monks.
Dave is the Perl 5.10 Pumpking, and (with the possible exception of chromatic) is the most knowledgeable and experienced Perl core developer to contribute to our online debate.
Dave seems to be frustrated by what he perceives as others (like me) just "not getting it", in reference to how difficult (or impossible) it may be to achieve significantly increased Perl 5 runtime performance.
Dave's technical views are clearly justified.
Dave may also be the most qualified person to guide us in the development of a Perl5-on-LLVM project, regardless of potential performance benefits.
chromatic (SW?) is a prominent Perl contributor and author, who is clearly quite familiar with the deep technical issues related to building a Perl 5 optimizing compiler.
chromatic's opinion seems to be filtered through a healthy does of skepticism, although perhaps he is a bit more willing than Dave to suspend his disbelief when given a conservative view of the LLVM project. (Maybe I'm just jumping to crazy conclusions again, hopefully I will be corrected if this is so.)
chromatic's technical views are clearly justified.
Like Dave, I believe chromatic possesses much of the skill needed to guide our initial development efforts. He has indicated his tentative willingness to be of some help in the future.
BrowserUK is a Pope here on Perl Monks, and an active Perl developer with a more-than-passing understanding of the Perl core.
BrowserUK has been the most vocal supporter of the possible benefits offered by LLVM, going so far as to begin compiling parts of the Perl core using LLVM and posting about the results.
BrowserUK's views on the future of LLVM and Perl 5 are clearly justified, even if his technical views are legitimately questioned by chromatic and Dave.
From private correspondence, it seems BrowserUK may have some feelings of regret surrounding the heated debate and potentially negative energy in the comments of my previous threads.
Can't We All Just Get Along?
I think all 3 guys are good, logical programmers who really do have the best intentions for Perl at heart. I feel bad that my threads are causing arguments, but at the same time I can see the possibility of fruitful ideas coming out of the fray once the dust has settled.
So, my question and sincere proposal is this:
Dave & chromatic, will you please continue to put up with the (perhaps ignorant) points-of-view of myself and BrowserUK?
Will you please not give up on us and the possibility of a better Perl 5 for everybody, even if not in the sense of a super-fast runtime I espouse?
Will you please use your powers to guide us and help us avoid traps that are obvious to only you?
Pretty please with cherries on top? ;-)
PS - I really do think that the refactoring required for Perl5-on-LLVM will benefit the Perl core developers at the very least, even if we fall short of the blinding speed I hope to achieve.