Hand-crafting a PDF

Question

Inspired by this answer, I thought an interesting challenge would be to golf a PDF file. PDF files are based on the PostScript language, so I thought this might be acceptable as a programming challenge for this site; but I wanted to check first.

A proposed challenge might be something like:

Construct the smallest possible PDF file consisting of a single A4 page with the text "Hello, world!" printed on it somewhere. The text may be in any font, but the characters must be at least 12 points in size.

Scoring will be by number of bytes in the file. External tools may be used to construct the file, but the score would be based on the size of the output PDF file itself, not on the size of the script used to generate it.

Entries must conform to the ISO 32000-1 PDF standard, and be viewable on commonly used PDF viewers.

Would such a challenge be on topic?

Answer 1

Your question seems on-topic, so I'll try to go over concerns I'd have about it being on topic, but say why I think it is on topic.

with the text "Hello, world!" printed on it somewhere

This is somewhat vague, does printed mean it has to be "text" or can I make an image and use that? I'm not certain how effective workarounds like this would be, but they might cause the challenge to degenerate somewhat.

This may mean your challenge would be too broad for having too many possible answers, but should be on topic.

based on the size of the output PDF file itself

This is usually fine, and it is what the tag code-challenge is for. However, Peter's comment addresses a potential problem with the way this is structured. Say I post and answer. Then, SomeUser copies the result that I got, but uses their massive server farm to compress it slightly. Now their score is better than mine. While there isn't really anything inherently wrong with this, and we've had several on-topic largest prime finding questions that function very similarly, I'd be worried that the challenge would quickly devolve into just spending more time compressing the current best answer.

Entries must conform to the ISO 32000-1 PDF standard, and be viewable on commonly used PDF viewers

This is a bit of a pain, but requiring conforming to a standard would mean you need to include the standard in your post. Instead of this, I would recommend changing this so that it only has to work on one free PDF viewer that is freely available and older than the challenge. This is more in line with how we handle programming languages as being valid for a challenge. (It also would help with the above, where finding lazier PDF viewers might help lower your score)

This may cause what you are asking to be unclear, but so long as you fully clarify what is going on your post is about writing a program to solve a task with an objective winning criterion and an objective validity criterion.

---

Some of this answer kind of crosses over to things that should be done in the sandbox, the core of your question: "Is this on topic" is yes, but the goodness of a challenge should probably be limited to the sandbox.

Answer 2

I think that such a question would be on topic, but would not be a good question for two reasons.

Firstly, as I commented, since the content of a PDF file can largely be DEFLATE-compressed, the battle for the last few bytes' saving would be a question of who has most CPU power to throw at kzip, zopfli, or whichever other nondeterministic extreme zip tool. This could be worked around by asking for PostScript instead of PDF.

However, that brings me to the second point. This is a single-language challenge, which are discouraged. If you want to promote golfing in printer languages, a better way would be to answer existing questions in those languages and optionally also to offer open bounties for answers which beat them.

Answer 3

I think you should try.

Yes, it seems like ontopic and how good is it can determine only the challenge itself.

Sandbox is useless as noone tried anything like that, so they can't help with possible problems and their solutions. The only moment is about conforming to standard. Let's replace it by it should be correctly displayed in Adobe Reader on windows or in [most popular viewer name here] on linux.

What's wrong with the sandbox in this situation?

How sandbox helps to improve the question? People are using their experience to predict and solve problems and potential holes of the challenge.

But what if there is a new type of challenge? Do the have any experience in it? No. Can they find a loopholes? Maybe they can think about it, but they wouldn't check. Now here are ideas about deflate, but will it really help to win? It's just assumptions - noone is going to chek.

So the right way is simply post the chellenge without trying to limit loopholes. Even if somebody will find and use them - it's not bad for the first time. And we'll get knowlege to disallow such ways in further challenges. Or we just see that noone is actually interested in such kind of golfing.

That's the only way that can show us real problems not imaginary ones.