LWN: Comments on "Development quotes of the week"

Development quotes of the week

chutzpah — Thu, 22 Apr 2021 01:35:11 +0000

After quite some time of using, and hacking on Linux boxes and getting by scripting in (what I now know to be somewhat rudimentary) shell, I decided to learn a "real" scripting language.

Just from the times I had to try to figure out an error by looking at the code, I selected Python. The main reason for the selection was the tendency of Perl code to make one's eyes glaze over, where generally you had some idea of what Python code was doing, even if you didn't know the language.

It's not much of a blue-vs-green spoon things as a spoon that was pretty easy to use and obvious how to use it versus a weird piece of cutlery that most of the world has no idea what to do with. For those that do know how it is used, it makes very specific types of soup way more delicious. Unfortunately other types of soup are rendered quite bland, sometimes bordering on inedible.

Development quotes of the week

smurf — Sat, 03 Apr 2021 22:03:12 +0000

Exactly.

For me personally I blame Perl's "TIMTOWTDI" motto for its demise, at least regarding my own Perl usage. "There is more than one way to do it." Yeah, right. When the time came to start using some object oriented ideas in our Perl codebase because, well, it became too unwieldy not to, there were too many ways to do it. Moose was just one of them, and none had any obvious "fit".

Contrast with Python – where everything's an object from the start, and writing metaclasses that build classes that mirror database records is almost absurdly easy, even if you're not using sqlalchemy.

I'd also like to know where in the programming language landscape Python had significant mindshare over Perl in the late 90s. I certainly don't remember any such places.

Development quotes of the week

smitty_one_each — Fri, 02 Apr 2021 23:12:05 +0000

I'd vary the "blame the community" point to understand that doing great stuff is only half the battle; advertising the great stuff done is the other half.

The python 2=>3 slog is instructive in that, unlike perl 5=>6, python was (eventually) successful.

Such transitions are Hofstadter's Law writ large, and perl's greybeards never got the marketing chemistry right.

Doesn't make them bad people.

Python's syntax has far less entropy going on, and the level lf effort required was still astounding.

Development quotes of the week

flussence — Thu, 01 Apr 2021 23:40:11 +0000

The problem with Perl has never been the colour of the spoons, but the social hazing rituals where they make you eat off a dirty floor for a month before you're allowed to own one.

Nothing was quite as eye-opening to me, perhaps about the rotten state of FOSS as a whole, as watching the escalatingly bitter harassment against (former) Perl 6 from Perl 5 elitists—perhaps after some of them realised they couldn't asset-flip their existing books in an afternoon for easy money. Once they'd finally driven the community away from their Old Boys' Club through constant heckling on multiple fronts, nearly daily, for well over a decade, all they have left is a language with most of the same architectural problems it had in *1999*, all of the reputation it had, and no more scapegoat to blame that on. It's turned into a production line of listless and confused thinkpieces like these trying to invent poetic new reasons to avoid taking responsibility.

Perl really internalised none of the lessons from the cups. Those who did left.

Development quotes of the week

wtanksleyjr — Thu, 01 Apr 2021 20:56:03 +0000

My reaction too. Perl was enormously popular, had far more modules, and was the standard language when you wanted to elevate from a batch file to a real script. Perl newsgroups and forums were far busier (and not just with helpless newbs).

I suspect that the problem was the huge gap when Perl6 spent so long in gestation (in vitro?). I don't know for sure, I know I wasn't doing any Perl during that time. I'd chosen Python a long time before, and all I worried about was the baffling-to-me 2->3 split.

Development quotes of the week

dvdeug — Thu, 01 Apr 2021 19:00:04 +0000

I think that "there's space for two strong languages in a small pool but when the pool becomes bigger there can be only one" is oversimplifying it, but the pattern can exist. If two languages split the market 60-40 or 70-30, both can be supported. If three languages split the market 60-20-20, it can be easier to support just the larger language than try to support three languages. When there's a number of languages, people will learn the language they have to learn (Python) and then pick the one that entices them most (Go, Rust, Elixir, Ruby, etc.) with way too many options in the second group to support them all.

In '98, it seems like there was Perl, Python, C, C++ and a few specialist languages. Sure, you could use Fortran, Pascal, etc., but they didn't really have any traction, and Java hadn't gained traction on Linux yet. Now, Java, C#, JavaScript, Python and Go all seem to have their ecosystems and vibrant communities. There's not a lot of space left for Perl.

Development quotes of the week

halla — Thu, 01 Apr 2021 18:17:13 +0000

That's my memory as well -- I got my impression mostly from articles in the Linux Journal, though.

Development quotes of the week

Karellen — Thu, 01 Apr 2021 17:39:10 +0000

"Python got an early mindshare advantage"... over Perl?

Really?

I always thought of Perl as having the early mindshare lead, at least through the mid-to-late '90s, and possibly into the early '00s. It was the sysadmin Swiss army chainsaw for when bash/awk/sed started to get unweildy in those days, with Python only starting to overtake mindshare in the early-to-mid '00s and after.

At least, that's how it seemed to me.

Development quotes of the week

smurf — Thu, 01 Apr 2021 13:31:31 +0000

For me it was Perl's brain load. You have "::" and "->", "$" and "@" and "%", $foo is distinct from @foo and/or %foo and/or foo except when it is not (and don't forget *foo), there's "my" and "our", there's a heap of magic variables … all of which has the potential to cause subtle bugs if you're paying attention to your algorithm instead of the Perl language itself.

Python? not so much. There's zero magic sigils, exactly one object/method/whatever accessor ".", and you indent your code anyway so why the hell would I want those stupid braces (and the finger acrobatics to access them on my non-English keyboard)? Thus I can focus on my algorithm instead. Also these days there's pylint and mypy which catch a heap of bugs for me. I have no idea whether an equivalent of these exists for Perl but I'm not even going to try find out.

So, no, this isn't a blue-vs-green spoon thing. Some of it, sure, but I strongly suspect that that's not the main reason.

Also, the "there's space for two strong languages in a small pool but when the pool becomes bigger there can be only one" is just … wrong IMHO.

Development quotes of the week

mgedmin — Thu, 01 Apr 2021 12:20:32 +0000

Perl was my favorite scripting language in the early 2000s. Then I had to learn Python for a new job. A couple of weeks later I started a hobby project in Perl, hacked on it for maybe 30 minutes, and realized that I'd rather use Python.

I don't think the reasons Python won were arbitrary. Perl had nice nifty syntastic shortcuts for processing text. Python made it easy to create abstractions.

Development quotes of the week

excors — Thu, 01 Apr 2021 09:23:51 +0000

The spoon argument largely matches my personal experience. Perl was one of the first programming languages I learned, and I really liked it, and I still occasionally use it nowadays. But when I started working more as part of a team where other people would need to read and modify and use my code, I had the perception that many people were unfamiliar with Perl and they'd be more comfortable if I used Python, so I used Python. The technical merits of the languages were largely irrelevant - neither is perfect but both are fine for similar types of programs. There just isn't enough room for two popular languages with so much overlap, and Python got an early mindshare advantage for fairly arbitrary reasons, so it ended up winning. That's not Perl's fault, so there's nothing Perl can fix that would let it regain a lead.

Slightly contradicting that, I also had the perception that Python was a more neutral language, not just a more widely-known one. Many people really hated Perl's syntax, and I didn't personally agree but I could see their point, while some people really hated Python's significant whitespace but not as many and I thought they were just being silly. And many people in the Perl community were really passionate about the language (putting intricate one-liners in their Usenet signatures, doing Perl Golf, writing poetry, writing the Acme joke modules that would e.g. turn your entire source code into executable whitespace, etc) whereas the Python community seemed pretty dull and humourless in comparison. When working with other people who aren't self-selected to be passionate about the same thing, dull is good, and I suspect that helped Python to spread more widely than Perl. That's still not much to do with the technical design of the language though, it's more about perceptions and about the community - and those things are extremely hard to change once they're established, and fixing the language itself won't help.