hpr-knowledge-base/hpr_transcripts/hpr4342.txt

Episode: 4342
Title: HPR4342: How I use Git to blog on the web and gopherspace
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr4342/hpr4342.mp3
Transcribed: 2025-10-25 23:21:59

---

This is Hacker Public Radio Episode 4342 for Tuesday the 25th of March 2025.
Today's show is entitled, How I Use Get to Blog on the Web and Go for Space.
It is hosted by Klau 2 and is about 38 minutes long.
It carries a clean flag.
The summary is, How I Use Get to Blog on the Web and Go for Space.
Hey everybody, this is Klau 2.
I'm going to talk about my unified solution for blogging on WWW as well as Go for.
Same blog posts, two different destinations, a single command.
It's a little bit of a hack, I'm not saying it's a beautiful solution, but then again,
I don't think anyone really accuses Go for of being beautiful at all, but this is working
for me pretty well.
For a while now, I've been blogging on mostly gaming related topics, a couple of tech articles
find their way through, but it's mostly gaming related stuff like tabletop, a lot of tabletop
stuff.
And I've been using a flat file kind of CMS so that I can post articles to this website
and have the, there's a PHP script that runs in the background.
I didn't write it, it's part of the CMS thing that takes mark down, which I write the
articles in and converts it to HTML and makes a nice landing page and kind of does all
the sort of the work that you would expect, I don't know, something like WordPress to do,
except instead of it happening sort of within your browser, it just happens in the background
of your server and the next minute and the next moment, you know, you go to your website
and there it is, there's everything.
So everything is pre sort of calculated, like the script converts the file that is required
and it puts it into the location that it's required to be placed into and so it's all done,
it ends in a flat file site system, but there's a bunch of scripts running to make all of that happen.
And the way that I interface with it personally is through Git.
This is not a grab approved method, this is something that I do because it's my preferred method
of working.
And I've been doing this, I don't know when the earliest post was, but I mean probably since
2023, maybe 2022, I don't really remember, but it's been a couple of years now and it's been
running quite smoothly, this is just the workflow that I've used.
It's a kind of a classic Git-based workflow.
I don't remember where exactly I actually learned it.
I feel like it was either from IBM DevWorks or I think DevWorks is the name of the old
blog that IBM, the tech blog that they used to have or it was from Bill Von Hagen, a boss that I
had at a company called Vivissimo, which was then bought by IBM. Anyway, it doesn't matter,
point is some really smart people told me about this little trick and it's essentially you use Git
hooks to sort of monitor on the server when something is received and that's a built-in feature
of Git. Like if you make a repository somewhere, get repository somewhere and you look in the hooks
directory or dot hooks, if it's a hidden directory, it kind of depends on whether it's a bare
repository or a normal repository. It's either called hooks or dot hooks. You look in that directory
and there's a bunch of sample. In fact, I can do it right now. I'll go to my temporary
directory. I'll make a temporary directory called my fake blog, my fake blog. I'll go into my
fake blog. I'll do git-dash, no, git and knit, right? Dash-bear I think and then a dot for here.
I don't know how much of that's actually, I think I think some of that wasn't necessary. But anyway,
I've just made a bare repository in my temporary folder and if I do an LS, there's a hooks directory.
If I go into hooks and do an LS, there's a bunch of scripts that end in dot sample. That's the
dot sample means, well, they're not going to run. Git doesn't run a script in the hooks directory that
ends in the word sample. I mean, I don't know. I haven't looked at the code to know for sure
that that's what. But to make it run, all you do is remove dot sample and then it would run.
And there's a bunch of scripts here that are pretty self-descriptive in titles. One is called,
or the one that I use most of all, is called post-receive. And I'm looking right now and I'm seeing
that that doesn't exist. I guess I made that one up. But it does work. If you do a post-receive script
in this directory, just trust me. It does run post-receive. I mean, I guess it's not, you know, the name,
of course, isn't really the thing controlling when the script runs. That's the point. These are
all examples of if you want to trigger something at a certain point in this repository's normal
operation, here you can put the scripts here in your hooks, your Git hooks directory. And when
something happens in that repository, these scripts will be checked to see if any of the triggers occur
and then the script can run. So, for instance, if you just look at, let's do a post-update. That's
similar, I guess, to post. That's not a very good one, actually. Let's do a pre-receive instead.
Pre-receive, yeah, that's a good one. So it kind of gives you some of the common variables and
things like that that you might encounter or have access to, I should say, within your Git
repository, when something occurs. So this example hook script, the pre-receive, echoes all push
options that start with echoback equals and rejects all pushes when the reject push option is
used to enable this hook, rename this file to pre-receive. So if you look at that kind of script,
you get access or you learn that you have access to things like Git underscore push
underscore option underscore count, which is kind of like an args or of the v-args and the
c-args kind of variable. It's kind of telling you, well, these are the options that were enabled
when this was pushed. So that's fine. So what I did on my server that receives Git content
pushes from my local machine. I created a hook, and I'll go into the hooks directory here,
and I'll do a cat on post receive. And in this one, it just does, again, now it's using little
built-in variables within Git that you kind of have to just read Git documentation to learn about,
but they do exist and it's things like while read old rev, that's the old revision,
space new rev, that's the new revision, and space ref name. So while we're reading all of those
values from a push that we have just received, do, and then the way I do it is I do b-r as in branch,
b-r equals dollar sign parentheses or backtick, whatever you forget, space rev dash parse dash dash
symbolic dash dash a brief or a brev dash ref space dollar sign ref name close backtick or
parentheses, whatever you prefer. So I've just identified the branch name with that incantation,
and that's one of those incantations where you don't even need to know really what it means. You just
look on the internet. How do I get the branch name from a git push, and it'll be right there.
Now if you don't want to just look on the internet for the answer, you can just read the docs for
git rev parse and find out which options filter everything down to the branch name, but either way,
you can assign a variable to what filters down to the branch name of the thing that was just pushed.
And again, the reason we have access to all this information is because we're using Git commands
within the Git system on data that that that because it's in the hooks folder, it's running against
this Git against this specific Git repository. So it knows its own sort of central, you know, directory
or it's what would it be the the base directory. Okay, so while we're doing that, oh yeah, we got
we got the branch name. So if the branch name dollar sign b r equals equals quote master closed master,
then WebDur, and this is a variable that I'm creating WebDur equals slash bar slash whatever it
is HTML slash, you know, my blog, whatever your blog fight, wherever your your your web directory is,
export GitDur equals dollar sign WebDur slash dot Git. And then push D. So we're switching over to
the WebDur's push D dollar sign WebDur Git pull pop D back to where we are and then finish the
operation. So if you follow it all that, then you know that we've just said if this push that we
just got was the master branch or main or whatever you call your your your top level branch,
then set the the WebDur to my HTML directory where I need all of my my files to be and then
export the GitDur to that WebDur directory is a hidden directory in that WebDur
we'll just call that dot Git. So that's a non bear repository. And then we switch over to that
directory to the WebDur. And then we do a Git pull. So we're we're switching over to a place that
we have not pushed to. So it doesn't have any of our personal private information or any access
or anything like that. This is just a clone of the stuff that I want on the web server. That's all
this is. Anyone could make this. You could make this yourself by just cloning my you know a
repository. It's not a big deal. This isn't this doesn't require that that Git directory doesn't have
any special access to anything. And then we do a Git pull within that directory. Now because I'm
using this Grave system, this CMS called Grave, I can I I have set up my Git repository to be
something that is compliant with what Grave expects to exist. And when Grave detects that there's
new files in those directories that I've just done a Git pull in, then it triggers a bunch of PHP
scripts in the background that transform my markdown files to HTML and applies all of the built-in
Grave themes and the ginger to templates and all the other things that it does and it makes the website
look pretty. So that's what I've been doing for for the HTML that the HTTP the WWW side of my
blog and it's been working quite well for several years. No problem. Again, the important note
here is that the the the the place where all of that happens like the place that the the the privilege
to Git pulling and pushing occur. That's not on the live web system. That's on the same server,
but it's not within like the web directory. Recently, I decided that I really wanted a space
outside of the WWW again. Been there before, but I never really felt like the traffic made the
effort worth it. And I think there is a bit of a balancing act there. I mean, it's kind of fun
for a while to post stuff on Go for or Gemini, but you do there there is the latent knowledge that
you are just kind of like publishing it to a place that probably people really, you know, most people
can't even get to well, I shouldn't say can't get to have not learned how to get to and and
and the people who who can get there who have learned how to get to access non HTTP content
over an internet connection. They just, you know, they they probably don't frequent your your site. So
sometimes you will sit around and kind of wonder, well, is that worth maintaining? And the answer
that I came to several years ago was no, it was not worth maintaining, but that was because I was
using a non unified system. I would have content here and then I would have content there and
sometimes the content would be the same and sometimes it would be different and then other times
it would start out the same, but then I'd make changes to one, but not the other. And it just it
literally, it just wasn't worth the effort. And so this time around in sort of my triumphant return
to go for such as it is, I decided that the way to make sure that I could post to both the
WWW and I'm saying the WWW very very point of the because it is the, you know, HTTP the protocol
is not sort of like my issue. I'm fine with HTTP. I'm fine with HTML. I actually quite like HTML
and CSS, but but the the the world wide web, the sort of the default street, the avenue,
the super, the information super highway that that is the the the internet just a there it's sort
of the charm of it has started to wane a little bit for me. So it's it's specifically WWW dot
that sort of that subdomain of the internet has just kind of lost some of its appeal lately for me.
I just feel like it's been there are a lot of sort of voices and influences on there that I'm
just not that fond of. I mean really really big corporations own it just seems like a bunch
of the websites and it just I don't feel like the sort of the voice of just kind of individuals
come through quite as strongly in the way that they used to. So I figure why not let's let's
start contributing to to the thing that I think is is a potential solution to this sort of this
loss of of kind of really culture. I think I think it's there there's a certain cultural sort of
component here that that that has faded from the internet. And I think that it persists not
in the same way. It's developed, but it it's that there's a certain kind of of
closeness or coziness on for instance go for or Gemini. So I thought well I should start posting
to go for because if there's no if there's no content there then people definitely won't go
there. I mean that's that's sort of a self fulfilling prophecy. So let's let's start posting
there. But how can I post there and on the the WWW how can I do both of those things in such a way
that I don't you know that it's invisible to me. Well turns out that my existing solution for
flat file CMS on the web could also work on go for and I'm I keep referencing Gemini sort of
in passing and that's because that is a target. It's an eventual target. I have not if not actually
started posting to Gemini because I'm not running a Gemini server on the server that I have access
to which is a server that deep geek and lost in Bronx and I share. So we are we're we're looking
at expanding our operations into Gemini but for now it's go for. So it turns out that that really
I mean almost obviously if you think about it the same solution can can totally work for for both
go for and WWW especially since I write all of my posts and mark down as it is which is really
just a plain text format. It has some fancy markup here and there I mean it's I guess it shouldn't
right it's it's mark down but I mean it does you know because there are things that you can't express
in pure text in some instances so it's got some specialized notation for like images which don't
really show up but you at least get to see what image you're missing or that you're not seeing an
image and there are you know URLs which currently I haven't like fixed or repaired you know such that
they point to internal go for links so it's it's a you know it's not it isn't like a perfect
solution I could I could get in there and and and do some do some cleanup if I wanted to but it's not
super urgent for me right now because I'm pretty happy with what I've got and and 50% of the work
I had already done for myself so once again it's just a single it's that little gets bait and switch
trick I create a get repository on the destination server and this is a bear repository so it's not
meant to be interacted with on the server it is a place that I push to that's all so I create a
bear repository and in the hooks directory post receive I do exactly the same script while read
old rev new rev ref name get rev pars symbolic a rev dash rev ref and then ref name look at the
ref name is that it does it match the branch that you want to to honor as the the primary branch
if so then set the web directory to some directory set your get directory to some sub directory
within that web directory and then push over to that web directory do a get pull and pop back
to where you are and then go through the cycle again so exactly the same script not I mean other
than in instead of setting your quote unquote web directory to slash vars slash html or slash
you know users share or not users share I don't know wherever people keep their html folders
these days I mean it could be in your home directory in public underscore html if that's the
kind of server you've got you know who knows point is you're you're putting it to var gofer slash
whatever my blog my fake blog or wherever you keep your gofer folder for you know wherever your
little gofer server has set your gofer files to exist in and then you do the get pull there and
now you've got to get directory that's populated by all of your just raw markdown posts now I wasn't
super happy I wasn't entirely happy with that and also I mean you can do like a directory listing
in gofer you can just tell it yeah like look at this directory and list all of the files like that
that's that's enough but the way that my CMS structures things is that everything is within a
sub directory and everything and all the files have a standardized name it's item.md so really
the title of the post is the directory I mean it isn't but I mean that's functionally that's
what happens is that your title your post title becomes the directory and then within that directory
there's the the the article content and that's just a peculiarity of the of the of the CMS I mean
if you were doing this on some other system for others some of the reason maybe that wouldn't be
the case for me that is the case so what I ended up doing was I wrote a quick little cron job
or a shell script rather which I do run as a cron job I think every just right I only
wrote run it once a day because I only anticipate ever posting one or scheduling one post a day
and that's an important component of this solution as well um one of the things that I want to be
able to do on my gofer server is schedule posts I don't want because I write posts months in advance
like I just I write a thought down in you know a hundred lines and I commit it to get and I
schedule it for a specific date and the grab CMS looks at the published date and the date lines
and for some reason it does apparently require both I don't know why but it looks at those lines
and figures out when it's supposed to actually appear on the website and then your post appears
on your website on the date that it's supposed to appear so it's got that scheduling component
which I quite like because yeah you could just dump everything on a server but I mean part of the
reason people go to a server you know part of the enjoyment I think of of blogs is that you go
every day and see if there's new content uh I post every other day so that that's kind of the
schedule so I wanted to be able to only have things up here on the day that I want them to appear
which of course is the uh the the I think either publish underscore date or just date uh value
in the in the post in the post content itself there's a header a yaml header at the top that you
can grip for so I just wrote a little uh copy post to blog shell script and it just it it does a
it establishes what day it is when this script is running and that's with a date dash dash rfc dash
3339 equals date and if you run that command you get uh something like you know 20 25 dash
o3 dash 10 let's say so there's that date and then I have it look in my uh base directory
which is where I have all of my new posts that I've posted to get so this is the get repository
that that that holds all of the um all of the posts no matter what date it is big people in the
world aren't seeing this this is the big master repository of a bunch of posts so I do have for
post in find uh my der base slasher staging type f name item dot md and then I exec grip dash hl
for the date that I've just established it is and assuming that the date the date in the post
matches two days date then we're in the loop and we do post der equals der name post so now that's
the the the the directory of the the post that I've just found right I'm finding the top I found
an item dot md file with the date in it so now I'm taking a look at the directory name that it's
contained in so now I have the name of the article and I copy the post to my base directory to my
live directory this is the one that the world does see but I name it I give it a name of the
the the the name is the the directories name so again the the title of the article is contained
in the directory and so I'm naming the post now the the text file I'm renaming it to whatever
directory it was found in so if I have a look at all my cool new toys directory right that's the name
of the post well that directory contains item dot md but now I'm I'm copying item dot md to the
live directory the one that the world sees and instead of calling it item dot md I'm calling it
look at all my cool new toys dot txt or something like that okay so I've just located a post it's
supposed to go live today I've copied it to my live directory I do one more step and this is a
very gofer I mean all of this is very specific but there's so when gofer like I say you can just
tell gofer hey look at that directory and give my user a list of the contents which is a great
option like I I really really quite like that it's one of those things that makes gofer actually
that's one of the things about gofer that is nice it's like hey look I just want to share a bunch of
files with you here are the files in a directory like you want to look at one open it up so I mean
I could I could you know with with this solution with a get solution and a a quick rename I mean
I wouldn't even have to do the rename to be honest that's really this whole cron job is a convenience
function like I could just say I could just do a get pull and and just point gofer to the resulting
directory the place where I've you know my live directory where I've copied all the posts to like
I could just be done with that but I do prefer for the posts themselves rather than being in sub
directories with the name item dot md I just think that's weird that's a weird non gofer like
experience so I do I do the rename that I've just described as a convenience for the readers but I
as a further convenience I guess for myself no no for the readers as well because okay so if you're
just looking at a big file of directories then if you're a reader and you come back to the sir you
yeah you come back to that gofer space and you think well that's you know I want to see if there's
a new blog you go to the big directory full of files and which one is new like what are you going to
do like a dip on the directory every day or something that's that's that's silly yeah I think the
value of just like here's a big directory full of files that's for for people who have discovered
the gofer space and just they they've got an evening and they want to read a bunch of stuff or they
just want kind of want to get a feel for what you're what you're offering here so it's you know
I don't feel like that experiences for like the return visitor like for the return visitor you want
that syndication that sense of syndication almost an RSS feed except I mean it's on gofer so there
is no RSS feed I mean I guess there could be in a weird way but I don't know how many RSS feed
readers take go for colon slash slash as a valid um you are I these days so I don't put the RSS
feed on the gofer server and instead I emulate it with this one weird trick which is that I create
a directory called by date or I think I call it date or something like that and the and and I'm
not saying I call it the date I'm calling I'm I call the directory d a t t e how do you spell date
d a t e yeah or you could call it you know by underscore date I mean it doesn't matter the
directory isn't the directory name isn't actually even gonna be seen by the readers point is
there's a directory and in this directory by date there's a file called gofer map and gofer map
is how you list how you manually list files on your gofer server it's kind of your index.html
for a lack of a better analogy I don't love gofer so don't don't take any of this like technically
speaking I do not love the way gofer manages things so don't take this as an endorsement I'm just
saying the gofer space is not the www and that to me is refreshing the the some of the conventions that
gofer uses I I just feel that they're inferior frankly but look a little bit of this a little bit
of that right I mean you know technology is perfect there's gonna be things that annoy me about
something so I mean this is this is the way gofer is it's got I got a folder and inside that
there there's this gofer map with an arcane syntax but again it's whatever it doesn't matter
right it's it's all it works so and this is cool so I do I I found a post with a date of today
that I want to add to my sort of my art my my non RSS feed so I do an echo dash e the dash e is a
it escapes some things because you do literally need a tab symbol like a tab character in your gofer
map like for it to parse correctly it has to be a tab so you have to do an echo dash e so I echo
dash e zero latest the zero is go for go for terminology for I think it's a text file or something
I'd have to look let me let me look I have I've got it's so non-sensical that I just have like a
go for cheat sheet in my home directory because I just how can you remember this yeah zero is a
text file all right so that's why it's zero latest because you're telling gofer that the thing
that you're listing here on this line of text is a text file so zero and then capital L latest so
this this text the the the name that gofer is going to provide to the reader is the word latest
they're not going to see the zero they the it's just gofer knows that it's a text file and it's
going to be called latest and then the single quote backslash T close single quote that of course
makes a tabbed character and then I do something like dot dot slash der live slash base name post
der dot txt and I echo that into a temporary file over in the temp directory then I do an echo
dash e of zero date and that is dollar sign date so this is the date that we've established it is
today space the results of bait or so you know backtick base name post der close back tip tick single
quote backslash or yeah backslash T close quote dot dot slash der live base name post der dot txt
and I echo that into my temporary files now I have two entries I've got one for the latest
blog post and then I've got one that actually lists the the current date it's the same article
listed twice I do it that way because I'm about to replace the previous latest entry with my new
latest entry so the latest entry is always getting replaced every other day every time there's a new
article and everything else all the the ones with the date are getting filtered are getting shifted
down a line and I just do that with a said command I said I find zero latest and I replace it with
the contents of my new file and then I remove the old latest line pretty pretty easy stuff
than I remove my temporary file and I'm done now this go for map listing these are just this
just points to locations so I'm not like duplicating posts one in the by date folder and one in the
by you know just the the big the big bucket folder like there there there it's there all all
one is a directory full of files that you actually see on the file system that's the big
by topic directories what I call it but it's just a it's it's a bucket of all the files while
the the folder called by date that contains just one file which is go for map and it is organized
or it is rather listing those files along with the date of the post which is the always the current
date when when this is being when the file is being modified so it's it's listing all of the files
by date and then pointing back to the actual files in the by topic directory this is not only
sort of like simple because again it's not you're not really there's the you're not duplicating files
you don't have to manage a bunch of stuff but it it also is quite effective for updates like
like updates and corrections and modifications so if I ever go back and change I don't know the
well if I if I just add something to a to an article or to a post that I've posted then when
I do a git push of course the git hook detects that I've pushed new con or that I've that yeah
that I've pushed new content to the master branch it copies the file to the live directory where
the live directory needs to be with the change in the the file name and everything and and no
modification is required for like the date updates like that that that listing is still correct
because it's always pointing back to this big bucket of all of the posts but the big bucket is
the thing being updated either every day when I or every other day when I post a thing or or
or rat well I don't I shouldn't say it that way because I don't post every other day that's
the point right I I'll write a bunch of things now in three days and then I'll forget about it
for two weeks and then I'll write a bunch of things for three days and then I'll forget about it for
two weeks that kind of thing so the updates just happen invisibly the the date the by date column
or page rather is just pointing to the the files whether they were the files with a mistake or
with a correction it doesn't care it's just a URI on the on the server so it's it it's a surprisingly
robust system that I that kind of took me by surprise I kind of felt like like when I made
updates I just thought to myself oh that's not going to it's not going to ripple through in the
go for server and I kind of felt bad because I was like I'm kind of back to a a dual head system
and this is kind of frustrating like a split brain kind of situation where I'm making updates to one
thing and it's not going to ripple through to the other thing but I guess I'll just have to I'll
just have to agree with myself that the get repository is the the central source of truth for
these things and you know if the the website or the go for site fall behind somehow then that's
just oh well but no actually it totally works like both of them are getting updates from get
once get gets those updates like the moment it's received in the bear get repository then the the
fresh version of that article is copied to the the public facing live you know live server and no
matter what like whether it's grab doing the intermediate processing of the markdown or whether
it's my a little cron job that echoes the name of an article by date whatever little system is
working in the background all it's ever doing is referencing a URI and that URI is is there no
matter what like that that that's the same URI no matter what I do with a get push that URI
exists I guess the one exception to that that I haven't really thought through is like oh my gosh
what if I publish an article and then I decide for some weird reason to change the name of the
directory that contains that file then I guess maybe that would be a little bit something that
that could cause a glitch in the system but at the same time I've done that in a total of zero times
since doing the blog so I I feel like if that ever comes up if I just start doing that for some
weird reason then maybe I'll I'll I'll have to sort of adjust make make adjustments to my little
system and because this is all centered around get it's really really easy to make this all happen
basically in one command I mean once you get the infrastructure set up I mean you you have to
certainly create your get pull or your get hooks rather on your server and do all the copying and
the any kind of cron job that you may or may not want or need you would have you know whatever
CMS you're running oh yeah the infrastructure is is flexible but the idea of triggering everything
with a get push is is you can you can you can target separate places with one command and and this
is really nice so you do a get remote this is on your local system that you get remote set dash
URL dash dash add dash dash push let's say um all ALL and then your you know URL your target for
it for the let's say your web server whatever that might be might be um you know your name at
my server dot or example dot com colon slash my get repository whatever uh and so now that exists
as a push destination for a remote called all that's one location so then you can do a get remote
set dash URL dash dash add dash dash push all and then another location your name at my go for
server dot com slash or colon slash um var slash go for slash blog dot get whatever and so now
you've got yet another push location so if you do a get push all ahead then now you're pushing all
of your changes to to both destinations with just one command from your local work work bench as
it were so really really easy to to make it trigger sort of chain reactions on two separate locations
or more locations if you're if you're you know eventually so I suppose I'll probably be pushing
to a Gemini server as well so who knows maybe that'll be a third destination probably not actually
that'll probably happen on the server but you get the point the idea is it's really really nice to
be able to just get everything sort of you know in one place one source of of of labor you're doing
all your work in one directory locally and then with just one get push command you're pushing it
to several different destinations the servers are doing their little jobs in the background and making
it available to your audience pretty pretty cool um I hope this is um maybe not taught you anything
new but maybe inspired you to use some new ideas uh and new hacks on the way that you uh manage
and distribute content thanks for listening you have been listening to hacker public radio
at hacker public radio does work today's show was contributed by a hbr listener like yourself
if you ever thought of recording podcast and click on our contribute link to find out how easy it
is hosting for hbr has been kindly provided by an honest host.com the internet archive and our
things.net unless otherwise stated today's show is released under creative commons attribution 4.0
international license