Hacker News

93% of paint splatters are valid Perl programs (2019)

482 points by - 59 comments

[3 hidden]5 mins ago

SIGBOVIK papers are almost always a delight

EDIT: From their 2026 publication

>Permission to make digital or hard copies of portions of this work for personal use is granted; permission to make digital or hard copies of portions of this work for classroom use is also granted, but seems ill-advised

[3 hidden]5 mins ago

Popular in:

2024 (515 points, 125 comments) https://news.ycombinator.com/item?id=40197013

2021 (431 points, 165 comments) https://news.ycombinator.com/item?id=27929730

[3 hidden]5 mins ago

And others! Here's the full list:

93% of paint splatters are valid Perl programs (2019) - https://news.ycombinator.com/item?id=40197013 - April 2024 (124 comments)

93% of Paint Splatters Are Valid Perl Programs (2019) - https://news.ycombinator.com/item?id=27929730 - July 2021 (163 comments)

93% of Paint Splatters Are Valid Perl Programs - https://news.ycombinator.com/item?id=20834351 - Aug 2019 (7 comments)

93% of Paint Splatters Are Valid Perl Programs - https://news.ycombinator.com/item?id=19576425 - April 2019 (132 comments)

[3 hidden]5 mins ago

You missed https://news.ycombinator.com/item?id=38754686 (1 comment)

(Found by clicking the ‘past’ hyperlink leads me to https://hn.algolia.com/?query=93%25%20of%20paint%20splatters..., which gives me six results)

More interestingly, gnabgib reported a different number of comments, leaking the fact that they have showdead on (and you have it off)

[3 hidden]5 mins ago

The reply near the top of the 2021 link with the Master Foo koan hits quite a bit differently in the age of vibecoding.

[3 hidden]5 mins ago

Tangentially related, the only way for a Java properties file to be invalid is if it contains an invalid Unicode escape sequence (“\u” not followed by four digits, and not preceded by an odd number of “\”). After resolving Unicode escapes, any character sequence is a valid properties file.

[3 hidden]5 mins ago

I've had a program where some R code stuffs some input in an URL, fires it at a server, then reads the returned CSV. No error handling or validation is attempted.

As it happens, basically anything is a valid CSV.

If the input is wrong, the server responds with an XML error. This got interpreted as a CSV. Every line became a record, but no column headers were found so every expected field is null/NA. All these nulls get nicely sorted and processed, which causes more string errors, etc ... Mayhem ensues when thst stuff ends up in a database.

[3 hidden]5 mins ago

CSV is not a format, its a vague idea, https://www.youtube.com/watch?v=YrqSp8m7fmk - this is a great video on how to make it fast, and how its pretty bad

[3 hidden]5 mins ago

Can't agree more. I've dealt with a CSV where multiple CSVs were cat'ted together, but the originals were not in the same dialect and differed even in charset encoding. Wrote a statistics based parser that decided which part of the file had to be decoded which way. Interesting technical challenge, but my god the reasons for its existence were dumb.

[3 hidden]5 mins ago

Falsehoods programmers believe about CSV: "Yeah, sure it's a mess and there are like, a gazillion different dialects of it, but at least once you know the dialect of a file, you can decode it"...

[3 hidden]5 mins ago

The actual paper is also highly entertaining - https://www.mcmillen.dev/sigbovik/2019.pdf

[3 hidden]5 mins ago

From the paper:

"An interesting thing about these images is that they’re not particularly Perl-specific; for example, the program shown in Figure −3 also evaluates as the number −3 in Python (and presumably several other programming languages.)"

We live in the age of Python, which I use every day. But I still also use Perl when it's the better tool for the job. And it often is!

[3 hidden]5 mins ago

In particular, I like "Figure − 3. − 3 ." as a good little easter egg.

[3 hidden]5 mins ago

  The remainder of the valid Perl programs are mostly valid due to a Perl
  language feature that is not commonly present in most programming languages.
  Namely, Perl has a feature called “unquoted strings”, in which a sequence
  of alphanumeric characters by itself is parsed as though it were a quoted string.
  As an example, Figure 5 is read by OCR as the text ME, which evaluates to the
  string “ME” even though the ME isn’t quoted. This would result in a syntax
  error in most other programming languages.

  (To be fair to Perl, when perl is run with the -w flag to enable warnings, it does
  helpfully inform the user that at some point in the future, the Perl developers
  will most likely pick gggijgziifiiffif as a new reserved word:
      Unquoted string "gggijgziifiiffif" may clash
      with future reserved word at - line 1.)

[3 hidden]5 mins ago

So they might be valid older(<v5?) PHP scripts as well?

[3 hidden]5 mins ago

PHP 5 had unquoted strings, and it looks like 7 did as well, but not 8.

[3 hidden]5 mins ago

For those who do not enjoy reading scientific papers this one is worth reading.

[3 hidden]5 mins ago

That's with the flexible perl5 parser with a dynamic lexer, which allows everything. For my new faster perl backend perl11-p2 I am using a PEG parser, similar to raku, which needs static rules for everything. So I bet it will not accept most of the noise. But I still can run only 50% of all perl programs.

[3 hidden]5 mins ago

I’m more interested in how the OCR missed that ‘Stitch & Shutter’ watermark in the third one.

[3 hidden]5 mins ago

It looks like the author used Tesseract, a traditional OCR image with lineage going back to HP flatbed scanners in the 1980s [1]

Obviously, it's been improved a lot since then - but traditional OCR is mostly about black text on white backgrounds. Or at least, the text being high-contrast relative to the rest of the document.

Light gold text on a white background, while there are black splatters? That's the sort of thing it would struggle with for sure.

[1] https://static.googleusercontent.com/media/research.google.c...

[3 hidden]5 mins ago

Hahahhahah good catch.

[3 hidden]5 mins ago

i loved programming perl. both the act, and the book. i long since gave it up, but i am still awed by how it could both be completely unceremonious and absurdly ceremonious. my $self = shift; for instance.

[3 hidden]5 mins ago

A whole new twist on "Hackers and Painters"

[3 hidden]5 mins ago

Someone should reverse that and see how many paintings can be made to execute as Perl and what those programs do when invoked.

[3 hidden]5 mins ago

You could kill two birds with one stone by entering them in the Obfuscated Perl Contest, were it still being held.

[3 hidden]5 mins ago

Every single day there's an Obfuscated Perl Contest being held somewhere on this planet....

[3 hidden]5 mins ago

> fruitful procrastination

It is so rare to find proper appreciation for this creative skill.

[3 hidden]5 mins ago

This is more about a bad OCR than about paint splatters. They also tried multiple OCR settings until one worked. If you used more accurate OCR this would be very different.

[3 hidden]5 mins ago

More accurate OCR would recognise the images have no text

  $ touch temp.pl; perl temp.pl; echo "exit code $?"
  exit code 0

which is a valid perl program. And that one image with the watermark "stitch & shutter" ?

  $ echo "stitch & shutter" > temp.pl; perl temp.pl; echo "exit code $?"
  exit code 0

Also valid perl.

[3 hidden]5 mins ago

It all depends on which OCR to use and therefore meaningless. It is especially more meaningless since they do not establish the baseline (OCR "pass" % on random input).

[3 hidden]5 mins ago

I think you’ve missed the point…

[3 hidden]5 mins ago

Honest question, and I’m not the person you answered to: what is the point? I sincerely don’t get it.

[3 hidden]5 mins ago

Well a) it’s mostly a joke, b) to the extent it’s not a joke, it’s about how Perl accepts absurd inputs as valid programs, c) OCR paint splatters are just a funny way to get to weird arbitrary strings, of which a strangely high percentage (say, 93%) are valid Perl programs, d) so the choice of OCR doesn’t really affect the “validity” of the result, and e) the author did actually try multiple OCR settings and, when multiple OCRs produced valid Perl programs, chose “the most “interesting.””

[3 hidden]5 mins ago

It's a joke about Perl. It's not meant to be taken seriously as a study about OCR.

[3 hidden]5 mins ago

The problem is the ocr, then.

[3 hidden]5 mins ago

Perl is such a ride. I wrote and maintained a nms/statistics/alerting stack around 2003-2004, ended up being around 200k LoC iirc. Just heard it is still in use at that company. My condolences to whoever has had to maintain that monster.

I was somewhat pleasantly surprised a few days ago when an LLM out of the blue decided to use perl, not python, for some simple scripting. There is hope in the world yet…

[3 hidden]5 mins ago

And Perl from 2003-2004 was especially bad. It took amazing effort to pass a file handle as a subroutine parameter. I once took a look at the Perl code behind Bugzilla, and quickly decided that it was better to let sleeping dogs lie.

[3 hidden]5 mins ago

93% of Perl Programs are valid Paint Spatters too.

[3 hidden]5 mins ago

Figure 6 represents the string “gggijgziifiiffif”, which by pure coincidence happens to accurately represent the authors’ verbal reaction upon learning that “unquoted strings” were a feature intentionally included in the Perl language

[3 hidden]5 mins ago

There is a gem in the footnote right below gggijgziifiiffif, where it says it's possible to make a perl error quine by writing this line to the right location:

    Illegal division by zero at /tmp/quine.pl line 1.

I just tried this perl 5.42.0, and any file name starting with a '/' will work.