IRC channel logs

<wizard>today i am going to rewrite my website for the fifth time in guile because it seems like fun :)

<wizard>how's everyone else doing

<jfred>Pretty good, I just did that recently - moved my blog to a Haunt static site haha

<jfred>I've rotated through several blogs over the years but I haven't had an actual personal website for a while. Hoping that having more than just a blog will help me stick to one thing in the long run XD

<cwebber>hi hi

<cwebber>oh nice wizard

<cwebber>are you going to use Haunt?

<cwebber>I love Haunt

<cwebber>I hear dthompson likes Haunt too ;)

<dthompson>yeah it's alright

<jfred>XD

<wizard>dunno what i really want to use yet, there seems like a lot of really good options out there

<wizard>might just use the built in web server, like what if i wanted to add some interactivity in there later :)

<dthompson>I'm partial to static sites for personal stuff because basically no risk of the web server falling over and/or getting owned

<wizard>yeah but the risk of falling over and getting owned is part of the thrill ;)

<wizard>but you're right, there's no reason to bring in functionality i'm realistically never going to need

<jfred>The more I've done sysadmin stuff for work, the less interest I have in doing more sysadmin stuff outside work haha

<jfred>so my website is on sourcehut pages now

<dthompson>jfred: this is *exactly* the way I feel

<wizard>the sysadmin part is the more fun part of it to me lol

<dthompson>used to do devops full time. it's work I can do but I don't want to do much of it if I'm not being paid lol

<wizard>i love it so long as i get to choose the tech stack

<dthompson>that definitely helps :)

<jfred>the sysadmin stuff was definitely more fun as a hobby before it was my full-time job

<dthompson>I've long been at the point where I wont self-host anything that requires a database server.

<jfred>this does also make me a bit hesitant to try and get people at work interested in guix. not that it's likely to work anyway, but I don't want to drain away my enjoyment of it XD

<dthompson>I can handle sqlite.

<dthompson>yeah I hear that, jfred

<jfred>I do run a Matrix server which is my biggest/most annoying thing to run

<jfred>can't really stop running that because by now I've gotten all my friends on it ^^;

<dthompson>I had a CS professor in college tell the class, who mostly wanted to do game development, something to the effect of: I like eating pizza, I enjoy making pizza at home sometimes, but I don't want to make pizzas for a living.

<dthompson>gotta be careful when monetizing something you enjoy

<jfred>yeah

<dthompson>at a past job I got guix into the pipeline of one particular specialized thing we did. it was replaced with docker when I was no longer part of the team that maintained it.

<jfred>dthompson: re: databases, I do enjoy things like sandstorm that hide the details of that from you as a user. like I'm sure some of the grains I have run mysql or something under the hood, but I don't have to care

<dthompson>as long as the abstraction doesn't leak and you don't have to go spelunking!

<jfred>indeed, haha

<dthompson>in my experience, eventually a database upgrade goes poorly and you have to decide if you have the energy to fix it or not

<jfred>sandstorm does a pretty decent job of that in my experience. other things I've used like yunohost have not

<dthompson>good on sandstorm :)

<dthompson>part of the reason I'm excited for more p2p stuff is less data to maintain on any given machine

<jfred>oh yeah, definitely

<jfred>I'm really interested in where some of the relay stuff will go with Spritely... if I could run some relay servers but most of the data is on end-user devices I'd be pretty happy

<dthompson>will be cool when we have an encrypted relay ready to go

<wizard>has anyone written an org mode parser for haunt?

<dthompson>no! it's a travesty

<wizard>darn!

<dthompson>org-mode is too complicated to write a feature-complete parser for but I have tried encouraging other people to make a library that parses a subset. would gladly integrate that into haunt.

<wizard>surprised nobody's written an org-mode parser for guile before, despite the complexity

<wizard>seems like a thing that would already have come from this community

<wizard>Maybe It Will Be Me (it won't)

<dthompson>there is sort of one

<dthompson>in guile-present

<dthompson>no one has ever extracted it into a standalone library that would be appropriate for haunt to support

<wizard>i also found this https://github.com/robert-lawrence/guile-orgfile

<dthompson>hmm not familiar with that one

<wizard>which seems almost pretty usable

<wizard>although unmaintained

<dthompson>wizard: https://gitlab.com/wingo/guile-present/-/blob/master/present/org-mode.scm?ref_type=heads

<wizard>oh interesting

<wizard>it seems like orgfile is just as complete as this even

<wizard>if not moreso

<cwebber>wizard: another option could be to use org-mode's s-expression exporter

<cwebber>and then have guile read that in

<cwebber>but also

<cwebber> https://orgmode.org/worg/dev/org-syntax.html

<cwebber>would be a good starting point to write your own

<cwebber>but try this

<cwebber>M-: (org-element-parse-buffer) <RET>

<dpk>but then you have a choice between using the Guile Elisp reader, and getting improper Scheme lists terminated by #nil, or using the Guile Scheme reader and just hoping the Elisp and Scheme s-expression syntaxes are similar enough, enough of the time

<dpk>you could also use the Pandoc Org mode reader

<cwebber>yes that's true dpk

<dpk>shell out to pandoc -f org -t json

<cwebber>lol

<cwebber>yeah that might be possible

<cwebber>I mean it should be

<cwebber>maybe haunt could have a general pandoc reader ;)

<cwebber>which uses the json export to read in all sorts of files ;)

<dpk>if there were a currently-viable binding of Pandoc to the C ABI that would be a much more attractive proposition

<dpk>there was one but i’m fairly sure it’s bitrotted significantly

<dpk>(i have done awful, awful things with pandoc -f org in the r7rs repo)

<wizard>many options!

<cwebber>fun fact

<cwebber>Heart of Spritely is an org-mode literate document

<cwebber>which is run for unit tests

<wizard>ooh i was actually looking at its source yesterday lol

<cwebber>it also does the wisp conversion stuff

<dpk>cwebber: oh, i plan to hook up the R7RS Large spec to something that ensures that all the examples actually evaluate in a real implementation (ideally multiple real implementations) to what they’re supposed to, at some point

<cwebber>I have a separate blogpost I really should finish that was meant to be a fun followup to the Scheme Primer

<cwebber>basically about church encoding etc

<cwebber>show how you can cut the metacircular evaluator down to just the lambda stuff and do any calculation

<dpk>more awful things to do with the worst markup system in the world

<cwebber>dpk: org-mode: there's nothing more comfortable, and nothing more organically grown

<cwebber>organic-mode

<dpk>org-mode is fine. the ungodly hybrid i’m currently writing the spec in – org-mode with a homegrown markup language i designed specifically for writing the entries in a Scheme specification – is an abomination which arose simply because i wanted to start writing the spec and not think too much about markup languages etc.

<dpk>so i just wrote some of it in Org and some of it in another markup language i already had lying around

<dpk>my plan to resolve this is … get rid of Org and move entirely to a custom markup language

<dpk>these are the fuckups we make on the way to being the first RnRS spec that’s designed to be read on the web first with a printout copy (PDF) as an afterthought, rather than the other way around

<wizard>"any sufficiently complicated org-mode document contains its own ad hoc, informally-specified, bug-ridden, slow implementation of half of Common Lisp"

<cwebber>lol

<dpk>other weird things you do while writing a Scheme report: today i installed PLT Scheme v208 from, i think, 2003

<dpk>i had to install 32 bit libc on my box first :D

<cwebber>oh wow

<cwebber>anyway there's only one good syntax for writing documents

<cwebber>and that's Skribe

<cwebber>ACTION ducks

<cwebber>of course civodul wouldn't fight me on that one ;)

<cwebber> https://www.nongnu.org/skribilo/ is fun if you've never looked at it

<dpk>i’m pretty sure we looked at everything

<cwebber>well not everyone has your apparently all-seeing abilities dpk

<cwebber>As a nerdsnipe, here's a paper about Skribe https://www-sop.inria.fr/members/Manuel.Serrano/publi/jfp05/article.html

<dpk>what we ended up with was the result of me deciding not to let a bikeshed argument about markup languages stop us from actually start work on producing something for people to read

<cwebber>basically what if text markup was quasiquotation

<cwebber>dpk: that's wise

<cwebber>Scribble in Racket is basically "what if that, but actually it was comfortable to write in"

<dpk>heh, yes

<cwebber>Scribble isn't quite as clean architecturally but it's very nice to author

<dpk>Scribble is really good! … but yeah, exactly

<cwebber>my wife wrote her dissertation in it

<dpk>it’s really strongly tied to Racket’s OOP system, is the main issue

<cwebber>yeah

<dpk>no way we could port it to any other implementation

<cwebber>I wrote an ODT exporter in Scribble

<cwebber>er, for scribble

<cwebber>for Morgan's dissertation

<cwebber>it worked pretty well! but was a mess

<cwebber>I spent some time reading org-mode's ODT code

<cwebber>it turns out that ODT is really good, actually

<cwebber>er ODF

<cwebber>with ODT just being one thing

<cwebber>it's just a zipfile with some xml files in it

<cwebber>if you've never tried unzipping a file saved from libreoffice, it's pretty interesting

<wizard>excel files were the same thing iirc

<wizard>word files*

<wizard>but probably excel files too

<cwebber>I thought word files were like a dump of the in-memory format but maybe I'm wrong

<dpk>old-school Word files were

<dpk>i.e. .doc

<dpk>.docx is the same idea as ODF

<cwebber>well that's because they were trying to kill ODF

<dpk>i was tooting with Conor Mc Bride last night and he was saying how much he’d like to get humanities people off .docx and into distributed version control

<dthompson>my current issue markup related issues center around markdown being very convenient a lot of the time and then being horribly unexpressive otherwise

<dthompson>current markup related issues* that is

<dpk>given i’ve had professors who *still* don’t use computers and get their teaching assistants to do everything that needs to be done on a computer for them … it’s going to be a while

<dpk>dthompson: yeah, that was why we didn’t choose Markdown despite people wanting a Markdown version for some reason

<dpk>probably the solution will be that i’ll write … a Pandoc JSON back-end for the parser for our custom markup language!

<dpk>then people can have whatever output format they like, probably with some degradation compared to a native backend for that format

<dpk>(did you know some people *like* GNU info??)

<dpk>but anyway, it does hurt when i write something in lovely plain text or Markdown or TeX and then i get it back from the editor for the first time and they’ve sent me a Word document version of it back

<dpk>although admittedly there’s probably nothing quite as good as Track Changes in Word/Pages/GoogleDocs for doing what it does. (maybe the editor could submit a pull request and comment on individual lines in the ‘review’ pane? not really ideal since they’d have to make all their proposed changes, *then* open a PR and add all their extra comments)

<dthompson>dpk: gnu info manuals *are* pleasant... in emacs. most efficient documentation system I've ever used, tbh.

<dthompson>the html export of texinfo documents is trash, though. just horrible.

<jfred>A custom reader for Skribilo that made it a bit more like Scribble would be neat (i.e. somewhere in between Skribilo's outline syntax and Skribe syntax)

<jfred>maybe that would be more complicated than I realized... looking at Scribble's documentation it looks like things like @section define an enclosing block that lasts until the next @section. so I guess a 1:1 mapping might be difficult :)

<jfred>(or something like that, anyway)

<dthompson>oh if I had infinite free time...

<jfred>:)

<jfred>I feel that. Many potential projects

<dthompson>when I made haunt nearly 10 years ago I should have also designed the perfect lispy markup language. what was I thinking??

<wizard>xkcd standards comic dot jpeg

<wizard>haunt review: i'm spending my time writing blog posts now instead of theming the website. this is a new experience for me.

<wizard>Can't Tell If I Like It!

<wizard>haha

IRC channel logs

2024-09-20.log