Help me resolve a domestic dispute issue about #twitter
Help me win a solid $50 by voting on the birdie site
https://twitter.com/ChristosArgyrop/status/1611533974345781250?t=M2jRKYot6xaN5Mp5ZXDNfg&s=19
ChristosArgyropoulos MD, PhD FlozinatorInChief on Twitter

“Place a bet with the wife about the longevity of twitter. I will come up with $50 if the site survives until the presidential election, and I will get $50 (hyperinflated) dollars if the site ceases to exist before the next elections. Are you with:”

Twitter
@ChristosArgyrop I think you need to define your terms, What do you mean by “survive”? E.g., #Myspace “survives,” but is a shadow of its former self.
@mjgardner www.twitter.com goes the way of the Atari. Wife thinks a MySpace future is possible, I claim clinical death.
@ChristosArgyrop @mjgardner

I think it is going to be like Yahoo Groups and fold. BlueSky might be able to create a user base about the size of the one for Diaspora from the remaining active users?

I wonder how many active accounts Twitter really has excluding the Bots, the trolls, and the general shit-posting accounts?
@HaplogroupNews @mjgardner it is really going downhill. I think Australia and NZ had an outage that lasted 14hrs? I would not hold my breath about Bluesky. I think setting one's server may be the only survivable option.
@ChristosArgyrop @mjgardner

Well, now that I have settled in a bit, I have to say that Misskey servers have impressive tools out the gate. I would never share one, but for a personal social media platform, they are great. Do you know TypeScript?

That is unless you have decided to join the Mastodon on EMACS crowd?
@HaplogroupNews @mjgardner not interested in TypeScript. My interest in computing is simply data crunching , pushing the envelope on the RNA sequencing protocols we are developing for research work (lots of text there, so I am back to #Perl ) and having fun after work. Never had an interest in anything related to web programming
@ChristosArgyrop @mjgardner

Ah yes. I have been having ChatGPT explain your papers to me.
@HaplogroupNews @ChristosArgyrop @mjgardner switching to the tech account . We have a nice short paper in preparation where we used #Perl 's regex to speed up a probability calculations for RNA sequencing applications. Too bad #Python's regex / re modules are not as fast. I have been telling a couple of friends who are in real AI/Machine Learning roles to revisit the "old" languages as they have constructs that can speed up their models
@ChristosArgyrop @HaplogroupNews @ChristosArgyrop You might even be able to get decent mileage out of #awk if you can fit your problem into how it expresses things.
@mjgardner @HaplogroupNews @ChristosArgyrop nah. Too many in the bioinformatics community show off their shell prowess and then complain about write only #perl code. This is the equivalent of many procedurally oriented physicians and I was never into this.
In more practical terms, I prefer having a more powerful scripting language at my disposal

@ChristosArgyrop @HaplogroupNews @ChristosArgyrop Oh, I agree completely. But I’ve been impressed with #awk’s raw performance compared to equivalent #Perl pipelines when I didn’t need to bring a full scripting language to bear on a problem.

In general, though, I’m with you—I tell the #bash-heads that Perl is waiting for them when their one-offs get bigger than a screen or two, and next time they might consider just starting with Perl.

@mjgardner @ChristosArgyrop @HaplogroupNews @ChristosArgyrop

Power of AWK is one of the reasons i was so late to the Perl party!
(P4/P5 transition; one would think P5 finally getting OO would have been important to me but i was still segregating ^scripting^ from ^programming^ then.)

@BRicker @mjgardner @ChristosArgyrop @HaplogroupNews I went the opposite direction. Way back when, I had developed a C program to do some calculations on images but did not want to write the stupid I/O interface and link the image libraries to analyze them through shell scripting. So I did it by using swig to wrap the C code and call it from perl using the ImageMagick library. The scripts still work , which is kinda surprising
@mjgardner @ChristosArgyrop @BRicker @HaplogroupNews glue is code. This was the amazing thing about #perl . It still is: the language delivered with the underpowered computers of the 90s; while we have more powerful ones, the tasks we are using them for, have grown even more. This is why I returned to the language : raw, unadulterated flexibility and considerable performance without compromise.
@Perl

@ChristosArgyrop @HaplogroupNews @ChristosArgyrop @mjgardner

Have you tried the re2 module for Python? It is much faster.

https://pypi.org/project/google-re2/
It's also available for other languages, including Perl.

Paper: https://swtch.com/~rsc/regexp/regexp3.html

google-re2

RE2 Python bindings

PyPI

@randomatic @ChristosArgyrop @HaplogroupNews @ChristosArgyrop A previous employer of mine used re::engine::RE2 for many things, because a) you can cap the memory usage to avoid #DoS attacks, and b) the lead developer was all about premature optimization. https://metacpan.org/pod/re::engine::RE2

But you can’t use the /x flag for better readability, and we ran into some nasty #Unicode bugs and had to fall back to regular Perl #RegExps in those cases.

re::engine::RE2 - RE2 regex engine - metacpan.org

RE2 regex engine

@mjgardner @randomatic @HaplogroupNews @ChristosArgyrop #Unicode support not critical for me, but limiting memory use is , as we are trying to run the codes in resource limited devices
@randomatic @HaplogroupNews @ChristosArgyrop @mjgardner I will take a look. We are using a particular regex that executes in our Xeons about 2900/s in #Perl , 10-20x slower in Python regex and more that 10,000 slower using re. Considering that our typical project has to run the regex between 1 to 20M times, and that there is competition for resources among the cores, speeding the regex would be invaluable (losing the /x flag sucks)

@ChristosArgyrop @randomatic @HaplogroupNews @ChristosArgyrop @mjgardner This always gets me interested in jumping in to do some profiling and comparison trying re in C, or whatnot language, for more tuned speedups.

But nothing I can remember beats Perl's existence already as a thin shim on C in easy, compact script.