Re: Blowfish Pseudocode

mm wrote:
In your opinion. To me the job is adequately accomplished with a
simple for loop.

Huh? You did it this way because you didn't know you could do it
otherwise.

I can think of at least three other "efficient" ways to solve the same
solution (one of which is faster than your binary search).

What's your point?

Your problem [aside from trolling] is that you're not willing to admit
you don't know what you're talking about. I suggest you actually go
out and measure the performance difference. Specially with things like
mp_exptmod() that don't even call the damn function.

Do you understand there is a difference between making something that
works and making something that works and, in more, that is well made?

I don't see tacking a binary search in there as "making it better."
It's just different is all.

You say that like it's a bad thing. This lis Lib*Tom*Math we're
talking about. It's my project and I can direct it any way I want.

"the way I can" would be less pretentious and, maybe, more appropriate.

No, want. At the end of the day they're my projects. Not yours. You
can fork them if you want, even rename it and take all the credit.
They're public domain. But so long as I am distributing them on
ltc.com they're my projects.

I reject maybe 1% of all suggestions that come my way for reasons of
correctness or suitability.

I may not have accepted your recommendation but at least that avenue is
open. I answer a half-dozen emails every week regarding support,
updates, upgrades, etc. You're not the first person to have a bug
report to file.

In fact, LTM is the "de facto" math library in the Tcl scripting
language ONLY because I worked with Kevin Kenny in making it suitable
(he found several bugs that I fixed).

The fact that you brought that up, as OT I might add, in an
inappropriate manner is because you're a troll.

It's normal to binary and search on something that takes less than one
iterations worth of cpu time? Not only is that more complicated [and
pron to errors] but it won't actualy make the typical usage pattern of
the library any more efficient.

No, it wouldn't be much more efficient. Just, people who have some
knowledge in multiprecision number programming would have less the
feeling that you pushed code without thinking.

You're right, I did write that without much thought to optimization.
Because it doesn't matter.

As I said numerous times, there are MANY places where optimizations can
go. Speed is not the goal of the LTM project. Efficient is. Clean is.
Portable is.

Binary searching something that takes less than one iteration of the
loop it saves is pointless. It doesn't speed up the code in any
significant manner and it makes the code harder to read.

In short, it doesn't serve the goals of LTM.

If your example of your technical know-how is micro-optimizing random
bits of code over the span of a year [been a few months since your last
"super optimization"] then it's good to know my competition is not
really that challenging.

They are not optimizations. That's how the work should be done.

In your mind. Like I said I can think of two other ways to implement
the function.

What I have to question is why you didn't just post this as a bug-fix
[or suggestion] like the rest of the people in my change log? What
makes you feel so threaten by my work that you have to go out and try
to harm my reputation publicly?

Are you joking? You are quite able to harm your reputation by yourself.

Really? By publicly addressing people like you?

Did you already forget that you published the logs of your downloaders
just to prove that you had downloaders (they could have lost their job
because of you)? Did you already forget you called "Barzatolski" the
author of VMPC just in order to belittle him? Did you already forget
your hundreds of "First of, *** off" (because you are so mature that
you are unable to manage the conflictual situations you create without
spitting insults)? And all your attacks against people like D. Bernstein
(you are the only one who doesn't understand that doing so is a kind of
suicide)? Your reputation? But you already made confetti with it a long
time ago.

Posting the logs is legal and I had every right to do that.

VMPC is crap, the author is a crank and I don't regret a single thing I
said to him.

DJB is a smart fellow, in abstract. I clash with him to bring
"reality" to the table.

I don't see how I have confetti for a reputation. So a handful of
cranks and trolls from usenet dislike me. Big deal. I've been to five
different countries so far where people know about my software and use
it. I know of people I'd from a dozen others. I can go to conferences
and have people I've never met know my name and my projects. My
website gets downloads not only from universities all over the planet
but from private sector groups. My code forms the backbone
[cryptographically at least] of several international endeavours such
as Tcl, Ruby, Dropbear and a few others.

In my first post, I just answered your "sarcasms" with other sarcasms.
Where is the problem? You have rights the others have not?

Your first reply you tore into LTM and my book projects, and called me
incompetent. That's not sarcasm, that's libel and there is a
difference.

You have yet to prove that not only is your solution of any speed
enhancement but that it's actually suitable for the well established
goals of the LTM project. I've stated the goals numerous times in
sci.crypt so you have no claim to ignorance on that front.

Is that you're jealous that I have put
together something that makes you feel inadequate?

ROTFL. Jealous of what? Not to be a member of the "We do'ers" club?

Just because you're not in the do'ers club doesn't mean you should mock
us.

Or is that you're
just anti-social and can't get along with others?

Coming from Tom "Me-myself-my-beautiful-self" StDenis, this is funny
enough.

It's funny that you're the only person saying this. You can claim I'm
selfish until you're blue in the face, but the nearly 50 others (at
last count) contributors to the LT projects and my literally millions
of end users [hint: video game industry] claim otherwise.

I know you're trolling and that what you say doesn't really amount to
jack squat but I find it hard to stomach how people like you can insist
I have this huge ego problem. I don't see you donating your time and
efforts to the public. Hypocracy much?

You're not the only one to find bugs [though in this case I don't think
you're right] in the LibTom projects.

Which bugs? We are talking about bad programming, not about bugs.

Troll.

But you are pretty much the only
one to have such a hard time expressing your views in a constructive
manner.

I never intended to help you, except the first time, a few years ago,
but you immediately normalized the situation : "Let me break it down for
ya sparky". You knew a little more than nothing and you were already
talking like the immense guru you will never be.

http://groups.google.com/group/sci.crypt/msg/87079f4033d103f7?dmode=source

You claimed I was ignorant because I didn't apply your
micro-optimization to the GCD function [collapsing two if statements
into one]. "Is that how you intend to answer at CRYPTO?"
[paraphrasing].

My response was that I wrote the function [among 100s of others] in a
relatively short amount of time and all I checked was whether the
function matched test vectors. I later explained that I intended to
fix the bug in due course [and I did].

To quote:

---
So when I say "made sense at the time" I mean when I wrote the function
it made sense because I had just spent 12 hours working on the library
[not to mention 16 hours the previous two weeks] and wrote it that way.
Recall I wrote the core of LibTomMath over the course of a few weeks
during my x-mas break [dec 2002]

I didn't say I was right nor was I arguing I was right. I said "it
made
sense at the time".
---

In short, you are an ***. You take everything as "oh look, I can
one up Tom" instead of "I can improve the project and help thousands of
people at the same time".

I invite bug fixes and suggestions, even from people like you. A bug
is a bug.

But just because you don't like my style doesn't mean I'll re-write the
project for you. If you have a legitimate gripe bring it up. If
you're trying to be petty and junevile I'll file you right next to the
other cranks in /dev/null.

Anyone makes a mistake
and it's "mm's little tyrade again!" It's a wonder if you're even
employed as a developer at all.

But you are not "anyone", Tommy. You are the one who wrote the "Fastest
fixed precision public domain crypto library in C and Assembler FOR FREE
over the world". And, overall, you are the one who insulted me a few
years ago. I know, you already forgot about that. But, don't worry,
I have enough memory for the both of us.

I don't see the basis for the hyperbole. TFM is

1. Fastest library on earth for the given platforms and problems
and
2. Is public domain.

Prove me otherwise and I'll retract the claim.

When you claim that I don't have the slighest clue about development or
performance I bring up that I do indeed know what I am talking about.
See, that's called "experience". Something you clearly don't [enough
of].

Experience is what tells us when and what to optimize and how to
present our concerns in rational and cogent fashions.

Tom

.