Matthias Valvekens

Oh wow, finally! ed25519 is an approved NIST digital signature standard: federalregister.gov/documents/2023…

People are already implementing ISO/TS 32001 and 32002 out there! (...and it's not just me, I promise!)

when you know what “statistically significant” means

@wtfpdf I think/hope we'll see some improvement once LaTeX ships with easy-to-use tagging support---and the academic publishing world gets around to updating their stuff... 🤞

@wtfpdf The LaTeX project is working on this exact problem: making it easier to produce well-tagged output from TeX with minimal user intervention.

@Oil_vampire @teleoqoelet @Atypical_Andy @harryhhutch For some applications, sure. But not always: there are lots of companies that make money by having industrial appliances shut down _voluntarily_ when power is scarce: #Industrial_customers" target="_blank" rel="nofollow noopener">en.wikipedia.org/wiki/Demand_re… These days, many grid operators have a programme like this.

@teleoqoelet @Atypical_Andy @harryhhutch Isn't true for industrial users. You can't only operate when energy is cheaper & plentiful. If you could it would call for increased capacity during those times & you have the same LCOE prob as with generation, destroying efficiency & overall productivity

The much-touted claim that “renewables are cheaper than fossil fuels” is based on the flawed concept of levelized cost (LCOE). Here’s a little analogy that explains why it’s BS: You’re heading out on Saturday night and you need a babysitter... 1/7🧵

@petervwyatt @wtfpdf Follow-up question: what would be the right colour profile for artwork on T-shirts? 🙈

@wtfpdf Did you submit your artwork as a PDF? 🤣

🥁🥁 You can now use EdDSA in #pdf signatures! ...well, at least we have an #ISO spec for it now. Quick write-up (+DIY demo) here: mvalvekens.be/blog/2022/iso-…

@MyMilkedEek ESTA? Anyway, as we all know, the famous list of misconceptions programmers hold about names applies _extra_ to US gov't contractors. 🙈

Well, thanks for telling me, US Government, but not much I can do about it.

@_plinth_ @MyMilkedEek By the way, this one is even more fun: 3 0 obj 3 0 R endobj ;)

@MatthiasVal @MyMilkedEek Recursion isn't an issue for something that is traversable in a loop. If it's not allowable, the soundest reason is this: 3 0 obj 4 0 R endobj 4 0 obj 3 0 R endobj Which is a circular reference, which is tricky to detect in the general case.

Can an Indirect Reference be indirect? I don't see anything in the #PDF specification, part 1 or 2, that would limit this. Not that it'd be _that_ useful, but still, wondering.

@_plinth_ @MyMilkedEek Sure, I should've been more nuanced w.r.t. the infinite loop <> stack overflow issue, but the effect is pretty much the same: you'd need to keep track of some state while traversing if you want to do this safely.

@_plinth_ @MyMilkedEek Notable exception: "naked" indirect objects are forbidden in object streams, but that's it.

@_plinth_ @MyMilkedEek Yeah, this issue is actually the subject of some current debate in the PDF standards community, actually. Blindly dereferencing recursively is problematic (if only because loops would blow up the stack), but there's indeed no clear file format requirement banning the practice...

While I use a MacBook when travelling, I've used a Linux box as my main dev machine for most of the past decade :). You're right that the "PDF on the Linux desktop" experience is lagging behind, though. TL;DR: I use arch btw

@jacko_os @wtfpdf Don't worry, that's a natural reaction 😁. Over-reliance on 2-byte wide encodings was a historical mistake (see also: the Windows API and Java). It made sense when there were <65536 codepoints, but that stopped being true decades ago, so... UTF-16 should really be obsolete. 🤷‍♂️

@MatthiasVal @wtfpdf Thanks both. Seems obvious once you said UTF-16, but I read that spec multiple times and each time I saw the word "unicode" and my mind replaced it with UTF8 🤦

Any PDF experts: Is "(\376\377\000H\000i)" a valid string, specifically for v1.2? It looks like a UTF8 BOM in octal, followed by a series of byte pairs of octal null+ASCII letter. Should a reader extract "Hi" from that? I've read the spec a few times and can't decide. @wtfpdf?

@wtfpdf @jacko_os As for PDF 1.2: this encoding scheme is described in section 4.4, although it's not yet called "UTF-16BE" in there (just "Unicode"), presumably because the whole "character set vs. encoding" debate was still ongoing at the time. But, TL;DR: Yep, this is also valid in PDF 1.2

@jacko_os @wtfpdf Yep, this is a perfectly valid PDF string. As a data type, PDF strings represent byte strings first and foremost, and this one maps to 0xFE 0xFF 0x00 0x48 0x00 0x69.

Friends don't let friends print and sign. #truefriendship #pdf