monkinetic the blog

Fedi/Mastodon programmers… with the #MastodonAPI, and given a url to a post on any instance (assuming I have access to the toot from my account), how might I get my instance to fetch it and give me a “local” ID that is suitable for passing as the “inReplyToID” in a toot payload?

Wondering if I need to:

• perform a search (https://docs.joinmastodon.org/methods/search/)
• find the relevant status in the results
• use the ID for the status

Would that be the “local” ID?

#MastodonAPI #fediverse #programming #blogging #indieweb

Over on Mastodon, @[email protected] said:

I’ll be excited to beta test another blogging platform. (Still a Tumblr nerd at heart.)

Unfortunately I’m not sure I’d inflict Goldfrog on anyone at this point. 🤣 However, it’s still fun to hack on, so here are some salient points:

My favorite features:

• A Twitter-sized posting form right at the top of the home page for posting short notes, supports Markdown, has a character counter for fun
• Posts support tags, and #hashtags in posts are auto-linked for searching by tag on the site
• I have a separate posting form I can use for longer posts
• Posts can be syndicated to Mastodon, POSSE-style

As for running Goldfrog, it’s just a Go binary running off a config file, a theme directory, and a directory of content. There’s a separate tool for indexing posts from the filesystem into sqlite (usually only done when installing or updating from git).

• Runs on a DO droplet, currently deployed with ansible
• The only “admin” functionality is posting and editing posts, everything else is editing a config in git. “Admin” is based on a security-through-obscurity login url and a hardcoded user and password in the configs. This is dumb ;) and I’m experimenting with supporting Gitlab OAUth for login.
• Posts are stored in git as Markdown files with yaml headers
• posts are read into a SQLite db for serving and searching

I enjoy hacking on Goldfrog, and sure I’d like it if someone had a barebones product like it, but I have no interest in supporting either an actual open-source product, or running an entire damn “platform”. 💜

What with Twitter (aka birdsite, hellsite, muskosite) flailing in the clammy hands of Dr. No, and interest in the federated web re-emerging, I figured it was time to review my own web presence and see what was the situation.

Dear reader, it was Not Good.

Warning one was hitting this site from my work network and getting a BitDefender screen of doom saying the site was serving a keylogger. NOT GOOD.

Then the site - which was hosted on Linode and runs my own homegrown blog software, Goldfrog - went completely down. After some “where did those ssh keys get to, where is this thing anyway” I got logged in and figured out that my server had been hacked in some way, TLS and letsencrypt removed. I haven’t had time to troll the logs for evidence as to how the server was accessed, but I downloaded them and have them set aside to look later.

We Can Rebuild It

Thus entered a week of figuring out once again how the heck Monkinetic is built and deployed, migrating the code from Github to Gitlab (which I’m more familiar with due to $dayjob), and refactoring the Ansible code that builds the server and deploys the blog/content.

Finally today I got it 85% done, which is pretty good for a full migration between hosting providers (I also moved from Linode to Digital Ocean where I already have some other services).

Masto-tootly-don

With the insanity on Twitter, I logged back into my Mastodon account on toot.cafe and enjoyed the huge stream of new folks migrating from Twitter to federated platforms (mostly to mastodon.social since that’s the first/largest instance, but folks are making their way from there to smaller instances as they get more comfortable).

Apparently Mastodon 4.0 is out (release candidate) and they’ve changed the annoying-until-it-was-gone “Toot” to “Publish”. I’d have preferred “Post” myself, but 🤷‍♀️.

Maybe servers should just change it to suit their audience?

I had no idea that the Apache Foundation had a blog app called Roller (Java, much engineered)

Now that I’m blogging more often, I often find myself uploading screenshots. MacOS’ capabilities are fine, except that I don’t really like uploading files with names like “Screen Shot 2020-02-10 at 10.38.34 PM”.

Yesterday I asked:

Are there any #macos screenshot utilities that will simply let me name the file before saving it? (My needs are simple, the OS’ own utility is quite sufficient except for the name thing)

My old Six Apart mate Simon Wistow mentioned Monosnap a free utility that does exactly that - lets me name the file before saving it.

(Conveniently, it also make recording gif screen captures stupid easy)

I’m sure it didn’t happen to everyone, but can you see when my use of Twitter took off?

I’ve imported my old tweets as Notes on the site here, which is why I can show those stats. To be fair, my blog had not been very active for a while, but it’s interesting to see over 20 years where my posting activity went:

If you follow me on social media - Twitter or Mastodon - you may have seen lots of nonsense posts go by recently…

I’ve been working on improving my POSSE features here, which meant not only composing my posts and notes locally, and then publishing them to Twitter et al, but also being able to track where they “landed” (ids and links) and make it easy for users to find my content on the syndicated site.

So I dug back into my syndication code, rewrote it several times, learned some things about goroutines, learned how not to do some things with goroutines, and settled on a way that worked. With luck, this post and any other that is also published on Twitter or Mastodon will have links to those sites along with the post, and (at least for Twitter right now) have links to reply, favorite, or retweet the post.

As Dave used to be fond of saying:

Still digging!

Chris Aldrich wrote a blog post about manually adding Webmentions for links to his posts from sites that are not themselves Webmention-enabled.

This reminded me that I’d like to add a “I linked to you” feature for the post detail page in #goldfrog for this site. (Goldfrog does support Webmentions, so Chris should get an automatic link from this post :))

For 2020, I’m writing a new blog app. It’s just for myself, a toy to remind me why I love the web. It’s called Goldfrog, and it sounds a bit like “Go, blog!”

Why in the hack, in this day and age, would I spend time writing my own #blogging software, when you can’t sign up for a VPS anywhere without tripping over offers to help you set up Wordpress, or Ghost, or what have you?

A few reasons.

New Year, New You

2019 was shite-filled, and due to politics, the tech trashfire, and the friction of blogging through several variations of static, git-powered versions of this site, I simply stopped blogging. I’ve wanted to, but the effort killed the motivation before I could get some words out.

So I finally decided to write something myself, that did just the things I wanted. #goldfrog is written in Go, because while I will love Python to my dying day, my brain needed a kick in the pants this year, which relates to my next point.

The Builder’s High

Rands writes eloquently on the builder’s high. With family engagements and work over the last few years my hobby coding has dropped to almost nil (None if I were writing Python).

I needed something to reboot my creative juices, and trying to write something I really wanted, that thought would be quick, in a new language, seemed like a good way to go (I did want it, it wasn’t easy, and Go hates me. But I’m learning and that feels great!)

Goldfrog

A bit more about Goldfrog: the single feature I wanted was a posting form on the home page, right up front. Various Userland products had had this, and it always felt right.

Second to this was an “Edit” link next to every post, wherever it was found on the site.

Finally, my main technical “innovation”: My content is still stored on the filesystem as Jekyll-compatible Markdown files. However, build times via Jekyll or Hugo are fairly slow for my 2800+ posts (since 2000, baby) and I hate that. So #goldfrog indexes all posts in a sqlite database on disk. Post creation and edits go to the DB and to the filesystem, so I can still periodically sync the changes to the git repo I have checked out there. But all the list views, archives, tag pages, and search functionality go to the DB, and are really fast.

The Setup

This is really the app I’ve wanted for a long time.

• It’s hosted on a Linode “nano”
• The app builds on CircleCI and the binary is pushed to an S3 bucket.
• I’ve got ansible playbooks that setup the VPS pretty much from scratch with Nginx fronting Goldfrog.

Next

I really need to get my logging story fixed, and I’ve got some idea on adding basic metric tracking to the app.

I’m writing a new blog app. It’s just for myself right now, a toy to remind me why I love the web.

It’s called Goldfrog, and it sounds a bit like “Go, blog!”