Replying to
@oraculix@ieji.de These are Type 8 pentagons https://commons.wikimedia.org/wiki/File:Pentagonal_tiling_type_8_animation.gif
Remote
Theo Honohan
@theohonohan@graphics.social
mastodon 4.7.3Interested in the roots of design
40 Followers
87 Following
40 Posts
Joined June 22, 2023
tumblr:
Replying to
@dpiponi@mathstodon.xyz Sinclair had a good line in graphic design, all right
https://www.computinghistory.org.uk/det/66310/Officially-Licenced-John-Harris-Sinclair-Artwork-Poster-Prints/
Replying to
@peter@mastodon.gassner.io I definitely see the beauty myself but I know people who wouldn't. I think your video is sophisticated.
It's like this Laufen marketing gif: a very brief loop that introduces organic variation in the shifting shadows of an off-screen tree. Irish people of my parents' generation would never have come up with this, it's too sensuous altogether.
Replying to
@peter@mastodon.gassner.io Lampson is trying to distill and pass on a lifetime of systems programming advice, and I definitely agree that the papers are not really practical cookbooks!
Replying to
@peter@mastodon.gassner.io OK so if I understand event sourcing, the basic idea is to use a log rather than just storing the state. Butler Lampson calls this architectural principle the being vs becoming distinction.
Since you have concerns about Westphal's thinking, I thought I would mention Lampson. I don't know if you are aware of his work – first, from 1983:
https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/acrobat-17.pdf and a recent update he wrote that might be more interesting and where he uses the being/becoming language:
https://arxiv.org/abs/2011.02455
Replying to
@dpiponi Most(?) people are only introduced to the formal "Euclid's division lemma" framing of division (with a unique quotient and remainder) at third level. But it's possible to express that in Prolog as well.
quotRem(N, D, Q, R) :-
leq(D, N),
leq(Q, N),
mul(D, Q, P),
add(P, R, N),
lt(R, D).
The symmetry of D and Q here feels like it has something to do with the 'partitive' vs 'quotative' interpretations of division, but maybe that's meaningless nonsense.
ChatGPT: "You have written code that relies on the identity (cos A + i sin A)^k = cos(kA) + i sin(kA), so I'm going to assume cos(kA) = cos(A)^k"
Replying to
@onpaperwings@typo.social The fact is that the Type Archive never managed to get itself "off the ground" in Kennington. It was not very easily accessible. I remember visiting around 2001 when there were plans to get the architects Marks Barfield to create a museum for the TA. It could well be that the outcome of all this is to make the collections more readily available.
I've recently started using jj to version the little coding experiments I work on.
Whenever I come back to the code I try running it, and then, if it works, use "jj new" to 'bank' that version. The risk that the working copy is in a broken state seems to be something that must be checked and that only good processes can avoid. Maybe I should be calling "jj new" as the last thing I do before walking away, but that seems counterintuitive.
"A more nuanced approach would differentiate between a function that just happens to be written in a different language, and one that actually performs an effect."
This is very suggestive to me. It reminds that you can have a procedure that uses mutable state, or even one in a really hairy imperative language, but have it provide a pure interface (as with Haskell's ST). A quicksort function, for example.
Replying to
@yaxu Typst is one current contender to replace TeX. Everyone agrees that TeX is a monstrosity. https://tex.stackexchange.com/questions/277114/what-is-texs-gullet
RE: @theohonohan@graphics.social
Another 1k employees gone, including at least one senior functional programming guy.
Deposit Return Schemes (for bottle recycling) are surprisingly tech-heavy. Instead of the RVM dispensing a physical token or charging up a card, it prints a simple paper voucher and informs a central database that it has done so. An EPOS then has to contact the same database when the voucher is redeemed.
An alternative system, "Digital DRS", puts unique codes on containers which are scanned by the consumer for instant credit (they must then recycle the container?) https://re-universe.com/our-projects/digital-drs/
Replying to
https://oeis.org/A007814 "Shows which bit to flip when creating the binary reflected Gray code (bits are numbered from the right, offset is 0). That is, A003188(n) XOR A003188(n+1) == 2^A007814(n). - Russ Cox, Dec 04 2010"
Replying to
@peter@mastodon.gassner.io Do you know the Eames film "Blacktop"? https://www.eamesoffice.com/the-work/blacktop/
Replying to
@drawkcab@hachyderm.io I don't know about that. The sed situation is essentially the existence of an "away colours" delimiter.
The TECO thing exists entirely to support terminals with no escape key by allowing the remapping of something else.
"Note that when an ESCape surrogate is set, the designated character (when typed at the console) is received by TECO as an ESCape -- that character is no longer directly available at the keyboard."
Replying to
@drawkcab@hachyderm.io sed has alternative delimiters and they're handy when you need to use the usual delimiter as a literal character. Maybe you are familiar with that.
The difference there might be that the rigid syntax takes care of defining the delimiter character, just from its position, whereas setting up TECO appears to be more complicated
https://stackoverflow.com/a/5864155
Replying to
@drawkcab@hachyderm.io I just remember unfortunate people in a CAD class I attended a few years ago here in Dublin (all of whom had some kind of disability) having to key in this cargo-cult double Return combo in AutoCAD.
The sed situation (being able to format a regex as 's#foo#fu#' instead of 's/foo/fu/', with no further configuration or fuss) is just a case where delimiters get out of the way.
Replying to
It's a lot less of an "interesting vacuous entity" when you look at it as Set, which of course has two values
Replying to
The distinction between an empty list and a list containing only an empty string is a bit like this advanced math scenario https://mathoverflow.net/a/46073
I suppose it only works in the setting where you're implicitly concatenating the strings, "" being the identity of that operation. I wish I knew more about this, but the obvious point is that a list of empty strings (how long?) contains more information than one empty string.
Replying to
@peter@mastodon.gassner.io It's not online, but there is an extract in https://youtu.be/gGPBX_6EhBo
What would be the closest analogue of Core War on the Harvard architecture? (https://en.wikipedia.org/wiki/Core_War)
Replying to
@int21h@mastodon.gal @MissingThePt@mastodon.social not necessarily. I think the point is that the words Sun and Moon should really be the same length for this to work. The diameters of Sun and Moon are ofc almost identical in the sky, whereas your version makes it look like the Moon is 1⅓ the diameter of the Sun and takes a long time to pass in front
Replying to
@peter@mastodon.gassner.io I think it's a bit transgressive to see beauty in (soap or detergent tainted) water on an asphalt surface! More so that rain on concrete.
Replying to
Just “people continuously create safety[1]” so it should be that "teams of programmers continually create consensus."
[1] https://how.complexsystems.fail/
If someone follows nobody on Github I just assume they're a raging egotist, same as I would if their social media profile looked like that.
Replying to
Another early discussion of side-effects (1971):
https://web.archive.org/web/20200907081014/https://www.clear.rice.edu/comp512/Lectures/Papers/1971-allen-catalog.pdf
This feature (custom emojis on the Fediverse) was news to me: https://codeberg.org/fediverse/fep/src/branch/main/fep/9098/fep-9098.md
The statement "the model was wearing an expensive dress" has an opaque reading, where the speaker is assumed to know nothing more about the dress. It also has a transparent reading, where the speaker is assumed to have a particular dress in mind. There's also a middle case where the speaker knows that the model tried on several expensive dresses, but doesn't know which one they were wearing at that moment. It's a restriction of the set of possibilities. https://people.umass.edu/partee/MGU_Web_13/materials/Handout-Kusliy_1up.pdf
Replying to
@dpiponi@mathstodon.xyz it's funny, i thought of Kan extensions when reading the first part of this. My understanding/example is that in a "tech tree" setting where morphisms describe which technologies depend on which, and everything maps to its year of invention, a left Kan extension to some new set of inventions will take a limit of the new invention's dependencies – which might look like a product (the set of years in which the prerequisites were invented), or the max of the dates if they're ordered.
Replying to
Replying to
"A Hamiltonian cycle can be constructed by moving across cube dimensions according to the transition sequence in a binary-reflected Gray code. Translating such a cycle to another source node by performing an exclusive-or operation on every node address by the source node index, assuming the first cycle has node zero as its source, creates 2^n = N paths."
Replying to
"These paths cannot be edge-disjoint. But, for all-to-all broadcast, there is no contention between data packets moving along the paths generated by a … Gray code. Moreover, it can be shown that paths generated by rotating the address bits in the Gray code can be used without contention, thus providing n paths for each node. The initial data set M in each node is divided into n packets of approximately equal size. Each subset is then exchanged with packets in all neighboring nodes in each step."