I'm an 👴 so I know C & Perl at an expert level. I'm solid in #Scheme, #Elisp, #Java, and #Python (but slow b/c I must look up details on standard APIs, etc.). I'm learning #Rust, knew #Ruby once upon a time, & dabbled in #Haskell. I also know #Tcl (of all things!).

I'm doing a medium text-processing/sqlite-db-update system. I'm still convinced that if you've gobs of wonky-formatted text flying at you, #Perl is still the right choice.

But I'm open-minded.

#ChangeMyMind #AskFedi #programming

@keithzg

This is very plausible theory. Despite Python being the perennial counter example used by haters, I just noted elsewhere in the thread I'm de-facto maintainer of a Python system that feels a bit "write once, read never" to me because it overuses esoteric Python language features.

There are dozens of ways (some well motivated!) a programmer can write software that no one (even themself) can maintain!

Cc: @cdonat @barubary

#programming #Perl #Python #polygot #Rust #Go

Fascinating analysis, @valhalla. I have spent my career (since I did graduate school in programming language internals) focused on the (false) interpreter/compiler dichotomy as a question of what the language *implementation* looked like.

I've never considered the question from essential a UI perspective — as your “REPL” test does. 🤔

@barubary's point that gdb could well be considered a REPL for C/C++ is also fascinating.

Cc: @gwadej @elricofmelnibone

#programming #Perl #Python #Java #Go

@lazyb0y

Esp. w/ LLM-gen-AI #slop, I've seen horrendously poor code written in every #programming language that I know.

I'm de-facto maintainer of a #Python system rn written that overuses so many esoteric features that folks who write Python *daily* reviewed & told me it's hard to maintain b/c it's “too clever for its own good”.

Every proglang has its own way to create “write once, read never” code. #Perl, C, & C++ do require extra human discipline to avoid that.

Cc: @cdonat @barubary

@barubary

This meme originated from #Java & #Python enthusiasts in late 1990s when #Perl's popularity outranked both of them.

I speculate the meme persists from non-polygots who are annoyed (or even insecure) that they can't fix code in any language that is presented to them.

Obvious answer to @cdonat's troll is Perl is one of a few languages that has effectively its own DSLs built in.

I remain fascinated at the need #programming snobs have to denigrate other people's work.🤷

#Go #polygot

@bkuhn @Sdowney

Hm, you want to get away from messy sh/awk/sed, but chose a language, that you yourself consider an almost-drop-in-replacement. How's that going to work out?

Back in the days I wrote medium sized applications in Perl: ~10k lines. I used to joke, that the only issue was, I couldn't read them. There is a reason, Perl is considered a "write-only-language" as soon as more than one developer is involved.

#Perl #Python #Go #programming

@gwadej

I did not go to the YAPC::NA that was in Houston. ISTR my last YAPC was in 2002, maybe?

#Perl will always be special to me as a FOSS activist because — while the 1992–2002 Perl community *definitely* had serious toxicity problems — I learned there what #FOSS community was about, and how #SoftwareFreedom could make #programming & usage of software egalitarian.

Fascinatingly, I later hung out in the #Python community, & it was much less software-freedom-oriented than Perl's community.

@gwadej

I have the same approach; I suspect we're the same age and era of computing so not surprising. But I generally stay really open-minded that new Perl code might be a Bad Idea™, which is why I started this thread.

I really don't think the small-to-medium system I'm working on this weekend would make sense to write in #Python or #Java. @paulv kinda did convinced me that *if* I had time to learn #Ruby well before the deliverable is due, it'd be a better choice.

Cc: @elricofmelnibone

@elricofmelnibone

I still use command-line compositions with cat and `[ef]?grep`, but anywhere `sed` or `awk` would go, I use `perl -[pn]e` instead.

I agree with awk++ analysis of Perl: it remains *at least that*, which is why I believe it has value. Some of the language-bias content in this thread willfully ignores this point: even if #Perl was once held up as a direct competitor to #Java, #Ruby, &/or #Python, there is also language evolution of purposes. (cf: #COBOL again).

#programming

@Sdowney

I think the fact that continued, current, regular usage sh/awk/sed indicates that there is value in Perl, *even if* only to show a more comprehensive & less messy way to do what is available in sh/awk/sed.

I find there is always a point in shell scripting where everything gets so much easier if you just make it a 250-line Perl script instead.

Python and Go *cannot* almost-drop-in-replace sh/awk/sed like that. I don't know any language but #Perl that can!

#Python #Go #programming

@hyblon
The joke is painfully apt. *Many* people use LLM-gen-AI to take software from their most hated programming language into their favorite — just 'cause. It's #disturbing.

I've tried this with (well written) #Perl code to #Python, & I definitely got #slop out when the work is human-unattended.

LLM-gen-AI can *help* a human do translations tasks, of course, but the temptation to “never look at code not in my favorite language since I have LLM-gen-AI” is dangerous.

Cc: @bignose

#LLM #AI

I like that:

if [condition]:
	breakpoint()

In other languages, I slightly dread starting a debug session because I have to remember the flag to pass extra args, add all the breakpoints on startup etc. This leads to long sessions because I don't want to lose the investment I've put into my breakpoints.

But in Python, I have much shorter debug sessions and I can get right to a point of invocation really quickly, verify something, and close the debugger straight away.

I use the repl much more because I can end up in a repl right with the context and variables that I need.

Defining more breakpoints within the repl is really nice to unpack nested datastructures in the same way it would be done in the code:

>>> for key, some_list in some_dict.items():
... 	for value in some_list:
... 		breakpoint()

It's really thoughtfully designed - thank you to the people who made it so!