ssh. https://mkws.sh is an OpenBSD VPS hosted by http://openbsd.amsterdam/, I do most of my work there. I run another small Void Linux VPS. I'm not that hardcore. π
https://mkws.sh is a OpenBSD machine. I'm not that hardcore. π
ssh. https://mkws.sh is an OpenBSD VPS hosted by http://openbsd.amsterdam/. I run another small Void Linux VPS. I'm not that hardcore. π
Win+;.
swallow effect. There's no need for text-mode graphics.
foo function call which is also not found because again there's no foo command anywhere, because the function definition wasn't parsed correctly earlier as preferred.
$ /bin/sh -c "$(cat foo); foo", the $(cat foo) part is evaluated in the outer sh process, so the actual argument your sh invocation is getting is:\n\n$ /bin/sh -c "foo() {\n printf "Hello World"\n}; foo"\n$ /bin/sh -c '$(cat foo); foo', the $(cat foo) part is evaluated in your sh process, so what's actually happening is $(cat foo) is trying to interpret the first "command" from the foo file, foo() which is obviously not found.
$ /bin/sh -c "$(cat foo); foo", the $(cat foo) part is evaluated in the outer sh process, so the actual argument your sh invocation is getting is:
$ /bin/sh -c "foo() {
printf "Hello World"
}; foo"
You have the function definition there and the call which works.
When calling
$ /bin/sh -c '$(cat foo); foo', the $(cat foo) part is evaluated in your sh process, so what's actually happening is $(cat foo) is trying to interpret the first "command" from the foo file, foo() which is obviously not found.
\n$ cat foo\nfoo() { printf "Hello World" }\n$ sh foo\nfoo[1]: syntax error: \n\n\n\nWorks well.\n\n\n\nDoesn't.\n\n I guess is how newlines are processed inside double quotes as opposed to single quotes.\n$ cat foo\nfoo() { printf "Hello World" }\n$ sh foo\nfoo[1]: syntax error: \n\n\n\nWorks well.\n\n\n\nDoesn't.\n\n I guess is how newlines are processed inside double quotes as opposed to single quotes.
$ cat foo
foo() { printf "Hello World" }
$ sh foo
foo[1]: syntax error:
$ cat foo
foo() {
printf "Hello World"
}
$ /bin/sh -c "$(cat foo); foo"
Works well.
/bin/sh -c '$(cat foo); foo'
Doesn't.
I guess is how newlines are processed inside double quotes as opposed to single quotes.seds on those html files and maybe create some temp files and pass those files to ./share/l.upphtml in the mkws main script?
t
f() {
uname
}
t2
#!/bin/sh
. ./t
f
ssh -T adi@REMOTE << EOF
$(cat t)
f
EOF
\nf() {\nuname\n}\n\n#!/bin/sh\n\n. ./t\nf\nssh -T adi@REMOTE << EOF\n$(cat t)\nf\nEOF\npp ? π Did you get pp: Buffer overflow? π
pp ? π
> After that, I tried to use a program bundled with someone else's shell script site generator to make my own, but I couldn't get around one of the absurd limitations of the original generator. I eventually conceded that I would have to drastically change the formatting of my website and continued working, and I very quickly ran into either a bug that I couldn't fix because I don't know C, or (more likely) an even more absurd limitation that I don't care to conform to.
Is this
pp ? π Did you get pp: Buffer overflow? π
sed 's/"/'\\''/g' I guess for double quotes.
sed 's/"/'\''/g' I guess for double quotes.
share/sitemap.uppxml:\n\n\n<?xml version='1.0' encoding='UTF-8'?>\n<urlset xmlns='http://www.sitemaps.org/schemas/sitemap/0.9'>\n#!\nfor f in *.html\ndo\n#!\n<url>\n <loc>$1/$(basename "$f")</loc>\n <lastmod>$(lmt -f '%Y-%m-%dT%H:%M:%SZ' "$f" | cut -d' ' -f1)</lastmod>\n <priority>1.0</priority>\n</url>\n#!\ndone\n#!\n</urlset>\nlmt is scripted in there, also in ./share/l.upphtml. You can script lmt also in your Atom Feed. pp is called all over the place.
share/sitemap.uppxml:
<?xml version='1.0' encoding='UTF-8'?>
<urlset xmlns='http://www.sitemaps.org/schemas/sitemap/0.9'>
#!
for f in *.html
do
#!
<url>
<loc>$1/$(basename "$f")</loc>
<lastmod>$(lmt -f '%Y-%m-%dT%H:%M:%SZ' "$f" | cut -d' ' -f1)</lastmod>
<priority>1.0</priority>
</url>
#!
done
#!
</urlset>
lmt is scripted in there, also in ./share/l.upphtml. You can script lmt also in your Atom Feed. pp is called all over the place.
share/sitemap.uppxml:\n\n\n<?xml version='1.0' encoding='UTF-8'?>\n<urlset xmlns='http://www.sitemaps.org/schemas/sitemap/0.9'>\n#!\nfind . -name '*.html' | while read f\ndo\n#!\n<url>\n <loc>$1/$(basename "$f" | pe)</loc>\n <lastmod>$(lmt -f '%Y-%m-%dT%H:%M:%SZ' "$f" | cut -d' ' -f1)</lastmod>\n <priority>1.0</priority>\n</url>\n#!\ndone\n#!\n</urlset>\nlmt is scripted in there, also in ./share/l.upphtml. You can script lmt also in your Atom Feed. pp is called all over the place.
mkws, the rest are dependencies. You don't run those manually and you can script them in your templates. Makes sense to call pp $anotherfile than singlebinary render $anotherfile I guess.
entr in my static site generator for instance, entr does its job very well. Also another thttpd or some other webserver or some weird websocket JavaScript live reload mechanism or another markdown processor when we have smu, cmark, lowdown, all of these with a lengthy set of switches and knobs on a single binary.
Markdown support in that single binary, and live reload, and a webserver. There's Hugo for that. Maybe Hugo is better.
wget -O - https://mkws.sh/mkws@4.0.8.tgz | tar -xzvf -.
pp and lmt. I really refuse to build another static site generator that has a webserver inside.
pp, lmt, mkws symlinks linking to one mkws main binary, or call one mkws main binary with the arguments mkws pp, mkws lmt, mkws main?. Or just build a single binary with arguments like "everybody" does it.
sh is a good fit as a templating language, it was built for massaging text, that's what templating languages mostly do.
tools, plugins or extensions section to the website. Also, you can check out some stuff from here also https://adi.tilde.institute/, linters and cl, cbl and fl which are CLI log "analytics" tools: unique, visits, referers etc. Might rewrite them in C not sure, but awk fits the bill pretty well. Some tinkering required for now.
Markdown capabilities to the main binary, or add a web server, or add live reload to the main binary. I kept it as simple as possible, I don't believe I really have to combine everything in a single executable, a tree is not bad.
Markdown capabilities to the main binary, or add a web server, or add live reaload to the main binary. I kept it as simple as possible, I don't believe I really have to combine everything in a single executable, a tree is not bad.
./share/l.upphtml file is https://clbin.com/phoub, where l.js is https://livejs.com/. This gives me a live reload env.
./bin/d script for development:
#!/bin/sh
export DEV=1
https &
find . -type f -name 'mkws' -o -name '*.upp*' | entr ./bin/mkws https://mkws.sh
https is https://clbin.com/tIIMk, entr is http://eradman.com/entrproject/.
./bin/d script for development:\n\n\n#!/bin/sh\n\nexport DEV=1\n\nhttps &\n\nfind . -type f -name 'mkws' -o -name '*.upp*' | entr ./bin/mkws https://mkws.sh\nhttps is https://clbin.com/tIIMk, entr is http://eradman.com/entrproject/.
mkws script for customization. pp, lmt and mkws could be combined in a single shell file via shar but that would give a complicated main mkws script I guess or combine them all in a single static binary but that would mean being unable to customize the mkws script. Anyway, give it a spin and let me know how it works.
pp and lmt I guess?
all is a broad scope. My idea was to distributed a full tree to and from which you can add/extract stuff to your preference. Read this also https://twtxt.net/conv/hvygjbq. There are also other static site generators who deliver a single static executable, hugo, zola, `saait`.
all is a broad scope. My idea was to distribute a full tree to and from which you can add/extract stuff to your preference. Read this also https://twtxt.net/conv/hvygjbq. There are also other static site generators who deliver a single static executable, hugo, zola, `saait`.
> but itβd be cool to have just one file static linked executable that does it all.
Maybe yes, maybe no. It would complicate some stuff, like customization, you would have to add config files,
all is a broad scope. My idea was to distribute a full tree to and from which you can add/extract stuff to your preference. Read this also https://twtxt.net/conv/hvygjbq. There are also other static site generators who deliver a single static executable, hugo, zola, `saait`.
sh file via https://man.openbsd.org/shar. I'll have to explore this.
uudecode not uuencode, this one:
ssh -T "$1" << EOF | uudecode
uudecode not uuencode, this one:\n\n\nssh -T "$1" << EOF | uudecode\n\ntar -czf - ws.sh|uuencode "\\$f"\n>&2 printf "Writing %s\\\\n" "\\$f"\n>&2 printf "Writing %s\\\\n" "\\$f" goes to stderr and doesn't interfere with uudecode which is getting it's input from stdout.
tar -czf - ws.sh|uuencode "\$f"
>&2 printf "Writing %s\\n" "\$f"
>&2 printf "Writing %s\\n" "\$f" goes to stderr and doesn't interfere with uuencode which is getting it's input from stdout.
\ntar -czf - ws.sh|uuencode "\\$f"\n>&2 printf "Writing %s\\\\n" "\\$f"\n>&2 printf "Writing %s\\\\n" "\\$f" goes to stderr and doesn't interfere with uuencode which is getting it's input from stdout.
4.0.9 is identical to 4.0.8. It's a bug.
\nssh -T "$1" << EOF | uudecode\n>&2 printf "Packing for %s\\\\n" "\\$(uname)"\ntrap "rm -rf $tmp" EXIT INT HUP TERM\ncd "$tmp"\n...\nf=mkws-"\\$(uname | tr '[:upper:]' '[:lower:]')"@"$(echo "$5"| tr _ .)".tgz\ntar -czf - ws.sh|uuencode "\\$f"\n>&2 printf "Writing %s\\\\n" "\\$f"\nEOF\n}\n
ssh -T "$1" << EOF | uudecode
>&2 printf "Packing for %s\\n" "\$(uname)"
trap "rm -rf $tmp" EXIT INT HUP TERM
cd "$tmp"
...
f=mkws-"\$(uname | tr '[:upper:]' '[:lower:]')"@"$(echo "$5"| tr _ .)".tgz
tar -czf - ws.sh|uuencode "\$f"
>&2 printf "Writing %s\\n" "\$f"
EOF
}
mkws's binaries on OpenBSD and Linux I run a set of commands via ssh, create a binary archive, pipe it to uuencode on the remote server and pipe it back to uudecode locally. Great use case for uuencode and uudecode pair.