IRC channel logs
2024-11-23.log
back to list of logs
<dokma>Is there a Guile implementation in Javascript? Would be nice to just run the same thing on the server and in the browser and for DB access. <sneek>I've been faithfully serving for one month and 22 days <sneek>This system has been up 17 weeks, 5 days, 1 hour, 54 minutes <marmalade>Anyone know why Guile would be ignoring my $GUILE_LOAD_PATH? <marmalade>lilyp: I mean, there's not much to tell. I set my GUILE_LOAD_PATH envvar immediately invoked Guile, typed %load-path and what I added to the env var just isn't there. <lilyp>perhaps a bash weirdness – did you use export? what if you set it on the same line? <marmalade>oh I'm using zsh. I set it in my .zshenv which is the top loaded dotfile afaik, checked using echo $GUILE_LOAD_PATH that it was there, then started guile <mwette>what shell; and how? you may need to `export GUILE_LOAD_PATH' <mwette>or `$ GUILE_LOAD_PATH=a:b:c guile' <marmalade>i don't understand if the env var is set it's set, no? <mwette>if you want shell executed commands to see variables you need to export <mwette>try this: $ foo=1 ; echo "foo=$foo"; bash -c 'echo "foo=$foo"' <marmalade>err, also i'm getting hardstuck trying to just install and use artanis <marmalade>as far as i can tell the guix repos don't have a glibc 2.8? <ArneBab>programming-basics-wisp#deploy (took me quite long to get guix pack to actually include binaries that find their library packages) <marmalade>this is intensely frustrating, the guile manual really should have a section that explains the usage of external modules <RavenJoad>Has anyone already modified the Haunt static website generator to generate valid HTML in the RSS/Atom builders? I want my site's RSS to include HTML in the <content:encoded> portion of the <item> so elfeed can render/highlight new posts when they go out. <mwette>marmalade: if you're on guix, did you try asking in #guix ? <RavenJoad>dthompson: Can Haunt generate CDATA-escaped HTML in RSS/Atom feeds? I want elfeed to see this HTML properly and do heading highlighting. <dthompson>RavenJoad: I don't think there's anything for that. I'm unfamiliar with what that does. <dthompson>if it's a correctness issue then let's patch it :) <RavenJoad>In RSS, the description is overloaded. By itself, it should be the content. But you can add <content:encoded> as a field to embed the post's content. You can then do <content:encoded><![CDATA[...]]></content:encoded> and put the normal XML/HTML in the dots. If the ... are HTML, then elfeed can recognize that and use the <h1> blocks to do minor highlighting of the post. <RavenJoad>This is not correctness, but would make reading Haunt-generated RSS posts nicer in the long run. <dthompson>okay so this is specifically rss or atom as well? <Arsen>CDATAs are generally an XML feature <dthompson>would it make sense to always use content:encoded or provide it as an option? <Arsen>it's kinda orthogonal to the description (kinda like summary vs real text) <RavenJoad>I have not fully investigated atom yet, but I image its content block (which is already marked as HTML by Haunt) is also escaped (< is replaced with <, etc.). <RavenJoad>I think providing it as an option would make more sense, though I think it should default to #t. It seems like everyone I follow via RSS puts the entirety of their post in <content:encoded>, so readers can read the whole post without needing to go to the site. <RavenJoad>dthompson: I have tracked the changes down to the sxml->xml step in (haunt builder rss). The hardest thing seems like teaching Guile's sxml how to handle CDATA. ssax:xml->sxml knows how to handle CDATA, but just removes it and returns the contained XML as a string. <dthompson>well ssax:xml->sxml is a parser so that doesn't seem important here <dthompson>need to see if sxml has a notation for cdata <dthompson>but I suppose it doesn't if the parser is stripping it out <dthompson>there's no way to do this without cdata? I find cdata really odd <Arsen>are you sure it must be CDATA? I see no reason for that <Arsen>it'd make the document significantly more hand-readable, so if a node could be marked as "use cdata please", that'd be nice, but in general not a bocker <RavenJoad>Agreed. I was toying with the CDATA version too, and was worried that would be a problem. If we can avoid that, that will make life a lot easier. <dthompson>if the sxml tools don't natively handle cdata then I'm inclined to not go out of my way <RavenJoad>re ssax:xml->sxml: The SSAX example was just to see if SXML had any way to represent CDATA. <RavenJoad>I'll toy with something and send a patch. We'll need to add the content namespace too. <RavenJoad>I'll put a patch together and send it in, so everything is done together. <dthompson>ah okay just another tag. great that's easy enough <mwette>the SXML spec should probably have used (*CDATA* "charsequence")