Annoying forum behaviour

[davieddy]

Originally Posted by xilman

I find it annoying that I'm forever being asked to slap down posters who appear either to be imbecilic or not potty-trained or both.

Je me regrette FA

David

[davieddy]

Originally Posted by davieddy

Je me regrette FA

David

First heard that ~50 years ago.
It still rocks.
Just noticed the punchline "Ca commence avec Toi"

David

[gd_barnes]

I find it annoying when people respond to themselves.

[Brian-E]

Originally Posted by gd_barnes

I find it annoying when people respond to themselves.

It could be worse. At least he's openly responding to himself. Another online community in which I play some role is plagued by a person who uses multiple profiles (sock puppets if you like) and responds to his own posts using different ones. Sometimes people do this to give their own viewpoints apparent backing and the link has some famous examples, but in this case his aim appears to be solely to annoy everyone else, lead them off what's supposed to be being discussed, and provoke angry reactions from the "real" people who haven't realised that we are dealing with just one individual.

If that ever happens here, I've never noticed it. So either the sock puppets here have me well and truly hoodwinked, or the gerbils do a good job at keeping them away.

[davieddy]

Originally Posted by gd_barnes

I find it annoying when people respond to themselves.

The 1 hour editing deadline is partly to blame.

[xilman]

Originally Posted by davieddy

The 1 hour editing deadline is partly to blame.

I've recently been accused of answering myself. The editing deadline for supermods appears to be infinite so that's not the reason in this case.

The real reason is that a significant time after the first post I found answers to my (implied) questions and so posted them so that others would be spared the labour of performing the search to find the answers.

[Christenson]

Originally Posted by xilman

I've recently been accused of answering myself. The editing deadline for supermods appears to be infinite so that's not the reason in this case.

The real reason is that a significant time after the first post I found answers to my (implied) questions and so posted them so that others would be spared the labour of performing the search to find the answers.

A re-post when you find your own answer isn't exactly the same as arguing with yourself...which is best done within one post, for example, if I had a heuristic argument that euler-mascheroni gamma wasn't transcendental...but didn't completely believe myseflf....

[gd_barnes]

Originally Posted by Christenson

...but didn't completely believe myseflf....

Now that is annoying. I find it very annoying when I don't completely believe myself.

[Flatlander]

Originally Posted by davieddy

Je me regrette FA

David

Looks like I'm too late. I was hoping to stop you.
I find it annoying when people try to turn every thread into a music thread to win a bet or something.

Also find it annoying when people insist on talking about things I don't understand. Like everything.

And when I ask a stupid question.

And when I ask two stupid questions and someone only answers one of them and I'm left wondering if I should repost.

When people start sentences with 'And'.

and er posts with little or no punctuation

[LiquidNitrogen]

Maybe not the place for this, since it applies to some of the software programs here and not posts in particular:

But if some of these apps are open source, and you have to run one app to sieve, then feed the output file as an input file to a prime prover app, it seems like the authors of each program should be able to "link up" their programs and make this a little more automatic.

OK, maybe I have been spoiled by coding too many cron tasks in my day to automate everything, but it seems to me it would be awesome if we could:

• Start generating primes by varying k's, b's, and n's in the k*b^n +/-1 field instead of varying just the k's
• Code a trigger point where after sieving slows down by some prescribed amount, the prime prover kicks in
• Continue the sieving on the sieving core, but "pull entries out" of the prime prover list if they have not been reached yet
• After some global condition is met, stop sieving, and generate more k's, b's, and n's and repeat the process

The idea is to constantly give the prime prover cores as much work as possible, yet improve efficiency by removing work when possible (deep sieving while prime proving).

Deep sieving prior to prime proving is a waste at some point, yet entering prime proving too soon is equally as counterproductive (an additional one hour of sieving could save days worth of unnecessary prime proving).

Why not have the best of both worlds? Start proving primes while sieving continues, and pull entries from the prover list as the siever finds more bad eggs.

[rogue]

Originally Posted by LiquidNitrogen

Maybe not the place for this, since it applies to some of the software programs here and not posts in particular:

But if some of these apps are open source, and you have to run one app to sieve, then feed the output file as an input file to a prime prover app, it seems like the authors of each program should be able to "link up" their programs and make this a little more automatic.

If it were only so easy on these long term projects. Some CPUs/GPUs are better at sieving. Others are better at PRPing. Most users are running multiple cores for their work and many projects have a vast combination of CPUs/GPUs to do the work. They target specific CPUs/GPUs with the work that they are best at doing.

While one might agree that linking srsieve to llr/pfgw would be nice, it is harder than you think. The developers of these software programs do not want to integrate their code to do multiple things, i.e. both sieving and PRPing. The other problem is that it makes one person responsible for optimizing both sides of the problem. Those who write sieves are very good at doing so, but don't know anything about PRPing.

For example, as maintainer of PFGW and phrot, if I were to integrate srsieve or sr2sieve into both, what do I do if someone comes along and writes a much better sieve? Now I would need to grab that code (if is was released as open source), strip out srsieve/sr2sieve, then put their code in its place. It isn't worth the effort.

That begs even more questions. There are multiple sieving programs, each target different problems. You have srsieve/sr2sieve, tpsieve, gcwsieve, fpsieve, and a number of other sieving programs. Although there are some similarities, the core algorithm within each is completely different. This gets even more complex when you consider if you are searching for a prime to attack a Riesel/Sierpinski conjecture or if you want to find a more generic Proth/Riesel prime. Even though the sieving software is the same, different approaches are used to choose the upper limit of n for the sieve. Sieving is not a "one size fits all" problem.

I'm not saying that it can't be done, just that it probably isn't worth the effort due to the number of variables involved. I suspect that most people are content with the current methods used because it gives them more granular control over how they want to search for primes.