What if I decided to use only web-based tools? I asked myself that question this morning, and spent much of the day finding out. I had a blast.

Tasks

I normally use Org-mode and OmniFocus for task management. Today, I tried two alternatives, Nirvana and TickTick.

Nirvana adheres closely to the Getting Things Done workflow, which I appreciate. It looks nice and has been around for a long time. People do complain about a lack of updates, but since I just started I don't really feel that yet. Nirvana seems a bit light on keyboard shortcuts, which is a shame. Overall, it feels similar enough to OmniFocus that I could adapt quickly.

TickTick is more like Things. It lets me organize tasks using lists and tags. Once things are scheduled, I can view everything upcoming. It goes way beyond Things, though. There are Kanban and Timeline views, a calendar, habit tracking, Pomodoro timers, etc. I feel like I've barely scratched the surface.

RSS Feed Reader

I have used NetNewsWire for years, but if I want to live in a browser, I need something else. There are tons of options. I decided to look at Feedbin, Feedly, and Inoreader.

Feedly was out immediately. At first glance, it looks nice, but it felt all wrong after that. I can't explain it, but we didn't click.

Feedbin is what most people I know recommended. I used Feedbin for a while years ago. It's good. It took too long to figure out how to tidy up the column showing the list of articles. By default it's filled with giant images, which means only a handful of articles fit on the screen. Once I fixed that, it was better.

Inoreader comes up a lot, but had never looked at it. Once I did, we clicked immediately. It just felt right, if that makes sense. It has a zillion features that I don't need, and I'm hoping I can filter out any "trending" or "recommended" nonsense. I signed up for a year.

Notes

This is funny. I pinned a Roam tab. It's unlikely this will last more than a hot minute, but man Roam was great. Still is, honestly. It's hard to justify using Roam these days, even though I still have 6 months remaining on my 5-year subscription. I have to admit that hammering notes into a Roam outline with super easy linking/backlinks/transclusion/etc. is still pretty cool.

Next?

This was only meant to a fun, very temporary experiment to kill a few hours this morning. It still is, but it's been so fun that I'll probably continue doing it tomorrow.

This is a bit esoteric, but I'm writing it down here anyway.

I have switched between using Denote and Org-roam for my Org-mode notes several times. They are mostly compatible, so this hasn't been too troublesome.

One thing I needed to do was make sure that all the .org files included an :ID: property at the top so that Org-roam includes them in its database. The ID property looks like this:

:PROPERTIES: :ID: 4def7046-3ef8-4436-a079-09362bf66aff :END:

Many of my Denote files already include this because I like to drag and drop files and images into the notes, and the ID property makes this work correctly. But there were maybe 300 files without an ID, and I had no desire to add the property manually to every one of them.

I started to write a shell script to do this, but got hung up on some bit of syntax, so I searched for the error and found the fix. Then I had trouble with the sed command, etc.

Sometimes I forget that ChatGPT exists. Sometimes I ignore it because I want to learn stuff on my own. But sometimes I just want the answer, and, problematic as LLMs can be, there's no denying their usefulness.

A couple of prompts, a few iterations, and I had what I needed in about 10 minutes.

#!/bin/bash

# Directory containing the Org-mode files
ORG_DIR="${1:-.}" # Defaults to current directory if not provided

# Loop through all .org files in the specified directory
for file in "$ORG_DIR"/*.org; do
	# Skip if no Org-mode files are found
	[ -e "$file" ] || continue

	# Check if the :ID: property exists in the first 5 lines
	if ! head -n 5 "$file" | grep -q "^:ID:"; then
		# Generate a unique ID (UUID)
		id=$(uuidgen)

		# Create the :PROPERTIES: block with the :ID:
		properties_block=":PROPERTIES:\n:ID: $id\n:END:"

		# Insert the :PROPERTIES: block at the very top of the file
		gsed -i "1i $properties_block" "$file"

		echo "Added :ID: $id inside a :PROPERTIES: block to $file"
	else
		echo ":ID: already exists in the first 5 lines of $file"
	fi
done

The only change I needed to make was to substitute  gsed for sedbecause the macOS version of sed was missing a switch and throwing an error.

While I was in there, I asked ChatGPT for a version of the script in Emacs lisp, just in case. I had it write the function for use in a Dired buffer, since that's where I'm most likely to want it. Here's the function...

(defun add-id-to-org-files-in-dired ()
  "Add :ID: property inside a :PROPERTIES: block at the top of all Org files in the current Dired buffer."
  (interactive)
  ;;(require 'uuidgen) ;; Ensure `uuidgen` is available
  (dired-map-dired-file-lines
   (lambda (file)
	 (when (and (string-suffix-p ".org" file t) (file-exists-p file))
	   (with-temp-buffer
		 (insert-file-contents file)
		 ;; Check the first 5 lines for an existing :ID: property
		 (unless (save-excursion
				   (goto-char (point-min))
				   (search-forward ":ID:" (line-end-position 5) t))
		   ;; Insert the :PROPERTIES: block at the top of the file
		   (goto-char (point-min))
		   (insert ":PROPERTIES:\n:ID: " (org-id-uuid) "\n:END:\n\n")
		   (write-region (point-min) (point-max) file))
		 (message "Added :ID: to %s" file))))))

Done and done. There may be better, cleaner ways to do this, but my problem is solved, which is all I wanted.

When reading one of my posts in Ghost, it would be nice if there was a quick way to get to the control panel to make edits. I always make typos, but don't always catch them until later. Or I might be reading, say, the /now page and decide to update it. By default, this involves going into the control panel and browsing or searching for the post/page in order to make the edits.

I found a solution to this on Jonas Liljegren's blog. Here's my slightly modified version of their solution.

Add the following to the post.hbs template:

    <div id="pi-pos">
      <a href="/ghost/#/editor/post/{{post.id}}">π</a>
    </div>

Add a similar one to the page.hbs template:

    <div id="pi-pos">
      <a href="/ghost/#/editor/page/{{page.id}}">π</a>
    </div>

Then add some custom CSS:

    /* Backdoor edit link */
    #pi-pos {
      position: fixed;
      bottom: 1em;
      right: 1em;
    } 
    #pi-pos a {
      position: absolute;
      bottom: -1.6px;
      right: 3.2px;
      text-decoration: none;
      color: black;
      opacity: .2;
    }

Now, at the bottom right of every post and page, there's a subtle π link that takes me directly to that content in Ghost's control panel. I've already used it several times this morning.

There is no built-in method in Ghost for floating an image and having text flow around it. It's a significant omission, and one which they say is "too hard" and have no plans to change. OK fine, I'll do it myself.

I found a reasonable solution in this post on the forums. Here's how I'm using it.

Add the following to the header in the code injection area:

/* small images to float but not look stupid on mobile */
@media (min-width: 40rem) {

.float-left-half figure,
.float-left-two-thirds figure {
    float: left;
    margin: 8px 20px 6px 0;
}
.float-right-half figure,
.float-right-two-thirds figure {
    float: right;
    margin: 8px 0 6px 20px;
}
.float-left-half figure,
.float-right-half figure {
    max-width: 50%;
}
@media (min-width: 64em) {
	.float-left-two-thirds figure,
	.float-right-two-thirds figure {
    	max-width: 67%;
    }
}

}

Then, in a post or page, I add an HTML block before the image card I want to float (e.g to the right).

Then close the </span> after the image card.

The Canon AE-1 Program was my first real camera. Unfortunately, I no longer own that original from 1982, so I bought another one in 2013. It's not a great camera, but it works. I like to put a roll through it now and then.

I spent the morning of the election with my daughter and grandson. It was a nice way to spend part of an otherwise anxiety-filled day.

OK, it's happening again. Hugo broke my site for the second time in two updates. I got mad (again) and decided it was time for us to break up.

I dusted off the version of the blog that I'd built using Eleventy and started working on getting everything updated. Except it didn't work. I don't know what I was missing or what had changed since I stopped using it, but things were broken. I then decided that I would start fresh with Eleventy's base blog repo. That was also a mistake. After several hours, I had built an ugly blog, without some of the features I'd wanted. I gave up.

In the meantime, the Hugo devs had fixed the bug that had broken things earlier. That meant I could stop what I was doing and get back to using Hugo and the template that I quite like. The problem was that I'd already decided I no longer wanted to use Hugo, even though it was working again.

So, I did something I told myself I would never do again; I spun up a Ghost instance. I have a complicated relationship with Ghost. It's fast, simple, and clean. Its editor is as good or better than everyone else's, and I can run an instance for under $5/month using PikaPods.

I was, as usual, immediately smitten with Ghost's editor. I made the usual tweaks to permalinks and other settings. I killed everything related to subscribing. Within an hour I had a nice, clean, easy blog that was ready to go. Except there was no content. I had an old backup from earlier this year, so I was able to import about 80 posts from that.

There are dozens of blog posts around that claim to have a solution for importing Hugo's Markdown files to Ghost's JSON format. None of them worked for me. I even asked ChatGPT for a script, and it was surprisingly good, but it imported pages and no posts.

Unless I can find a way to automate the import, I'm left with the task of manually moving about 250 posts. It's almost enough to put me off the whole affair, but I'm typing this in a browser on my laptop in bed, and I'm not worried about Hugo versions or sync status or SSH keys or any of that. It's pretty compelling, so I'm planning to move ahead with the migration.

Oh, I almost forgot the pros and cons.

Pros

• Great editor: Images, callouts, linking, fancy blockquotes. It's very nice.
• Can post from anywhere with a working browser
• Fast
• Reasonably cheap to run
• Decent selection of themes and it's simple to switch between them
• I can send emails out for new posts if I choose
• They are working on ActivityPub integration
• It doesn't break every 5 minutes

Cons

• It's not a static website
• While I could edit posts and somehow post them using Emacs, it's not really worth it.
• Themes are expensive
• I lose some nerd cred
• They're focused on "monetizing your audience". Gross.

Some quick notes on my move back to Doom Emacs after a few days.

After once again hitching my wagon to Doom Emacs, I have been both elated and frustrated. I'm elated because Doom adds so many nice little quality-of-life improvements that make using Emacs downright pleasant right out of the gate. On the other hand, it ruins a few things and sometimes breaks for no reason that I can understand. It's nice having someone maintain the basics of my config, but it's also frustrating when those someones do it "wrong". And sometimes running ./bin/doom upgrade breaks things. I can usually recover, but don't love that I have to think about it. This is a side effect of having others do things for me, and it's a fair trade.

So far, I've resolved most of the little issues. One of those was with Elfeed. Elfeed is unusable for me when using Doom's default configuration. Doom includes the elfeed-extras package, and I don't like it. The split window is annoying. What's worse is that there's no date column in the list of articles and there's no simple way to include it. That's just dumb, imo. So I disable that package and modify a few little things and it's much better.

The remaining problem is that I sync my entire .config/emacs and .config/doom directories, and this somehow breaks because Doom adds $TMPDIR to the .config/emacs/.local/env file. Apparently, my tmp directory is not the same on both the Mini and the MBP, so I get permissions errors every time I switch machines. The workaround is to run ./bin/doom env before starting emacs when switching machines. That's not sustainable. I'll figure it out, but it's one thing that still bugs me.

And oh, the key bindings! Around six months ago, I moved back to my vanilla config and decided to stop using Evil mode. It was a painful transition, but I got used to it and now the stock key bindings feel normal. The problem was that I also use several tools that only offer Vim bindings. Switching between Emacs and Vim bindings has been chaotic, to say the least. I keep tripping over myself and there's been a lot of swearing. Going back to Doom and Evil mode has been tricky, but the muscle memory is returning, and I like the consistency in the apps I use most.

Something I dislike is using Doom's abstractions like after! and map!. It just makes things even less "normal". Handy, but it will make moving out of Doom harder. Not that I'd ever do that, though, right? 😆.

Right now, I'm happy with the setup. I love when Doom does something and it makes me say, "ooh, nice!". As long as that happens more often than me saying, "WTF?! That's dumb.", I should be fine in Doom.

I've decided to dive back into Org-roam and see how it feels. I've done this before, but eventually moved back to Denote. I love Denote's simplicity, but I have noticed that I don't take advantage of many of its main advantages. For example, I always use Org-mode files, never Markdown or plain .txt. I almost never filter my files based solely on components of their names. I like using multiple tags, but doing so makes Denote filenames even more unweildy. And so on.

How did I end up back in Org-roam? I summed it up yesterday, so start there.

I'm getting over my internet-induced fear of "lock-in". I mean, Org-roam uses a (GASP!) database! Who in their right mind relies on a database for managing their precious plain-text notes? 🙄🙋‍♂️. You can call it lock-in if you want, but the first requirement for my notes is that they are as useful as possible to me...today. If that means there's a chance they end up less useful in some hypothetical scenario 20 years from now, I don't care. Besides, they will be useful in 20 years. They're still just plain text. The database is only a wrapper around a bunch of text files, so without that the only important thing I lose is the convenient linking and backlinks, etc. So they'll just be a little less useful, I suppose. That's fine, and a trade-off I'm willing to make.

One thing I'm doing to help with future-proofing is to use org-roam-dblocks from Chris Barrett. This lets me "burn" backlinks or query results directly into the files, removing the need for the database-ey bits. Here's an example:

(use-package org-roam-dblocks
  :after org-roam
  :load-path "~/.config/emacs/lisp/nursery/lisp"
  :hook (org-mode . org-roam-dblocks-autoupdate-mode))

I have it configured to automatically update whenever the file is opened/saved. That helps. Without this, though, there's always search 😀.

Speaking of search, I've configured the wonderful Deadgrep as the default way I search in Org-roam. It required that I write a small lisp function, but it works great. I love Deadgrep.

(defun jab/search-roam (search-term dir)
      "deadgrep in org-roam-directory"
      (interactive (let ((search-term (deadgrep--read-search-term)))
                     (list search-term org-roam-directory)))
      (deadgrep search-term dir))

Here's what the Deadgrep search results look like:

That earlier post explained why I like using Org-roam, but here's a quick list:

• Any org heading can be a node
• Aliases
• Multiple tags without cluttering up the file names
• Feature-rich Daily notes functionality
• Org-roam-refile

I even like the Graph made possible by org-roam-ui, even though I make fun of it.

I am hedging my bets just a little, though. My default org-roam-capture-template creates new files using Denote's default format. It also includes the #+identifer: front matter that Denote uses. While I don't intend to use both Denote and Org-roam like I used to, it's nice to know that I can.

I sure wish Hugo would stop deprecating things.

This time it's site.Social. My theme uses that in a couple of files, and the theme's author is busy and doesn't always have time to fix things right away.

They now want site.Params.Social so I've temporarily fixed the problem myself by editing two files.

In twitter_cards.html:

{{ with site.Social.twitter -}}
# becomes
{{ with .Site.Params.Social.twitter -}}

And in opengraph.html:

{{- with site.Social.facebook_admin }}
# becomes
{{- with .Site.Params.Social.facebook_admin }}

At least now the error is gone and the site builds. I'll need to keep an eye on the PaperMod repo for a real fix.

UPDATE: See Issue #1573