cjkarr.bsky.social
I wear a lot of hats: software developer, entrepreneur, weird fiction publisher, writer, non-profit founder, election worker, local handyman, open-source creator, and Macross fan.
Latest updates can be found at notesfromthevoid.cc.
2,644 posts
552 followers
684 following
Regular Contributor
Active Commenter
comment in response to
post
comment in response to
post
Grok is like Caprica Six I think
comment in response to
post
My condolences.
comment in response to
post
"Wonder what was the last thing was that went through his head?"
"A missile."
comment in response to
post
Your dog argues that you lacked standing to file the complaint in the first case.
comment in response to
post
The Javier Saltares and Mark Texeira era was one that will never be surpassed.
Perfect art styles for perfectly dark stories.
comment in response to
post
They're actually building their own Operation Jade Helm.
comment in response to
post
... as opposed to putting a foot on their heads to remind them that they're only here because kinder Americans in the past signed treaties that allowed their entry after sending a boatload of Jews back to Hitler to be slaughtered 85 years ago.
comment in response to
post
The idea is that it would be something like an immigrant Head Start program where folks could stay while finding opportunities for their skills, or pick up skills in demand to fill labor holes that the existing population was neglecting.
The intention was to give people a leg up as new Americans...
comment in response to
post
To put my cards on the table, in the past, I've advocated for staging cities where folks coming in can be directed to a place to receive language instruction, job training, and legal services to give them a leg up before entering "regular" America, but this isn't that.
comment in response to
post
The outage last night wasn't great, but this is a lot of new code and infrastructure - it recovered on its own and there are MANY worse ways it could have unfolded.
On that note, back to work on a mobile app (and supporting infrastructure) for helping teens take their cancer medication.
comment in response to
post
On the plus side, I wasn't flooded with server mails this morning - just the ones I needed to see.
No more waking up to hundreds of these messages to sort through daily.
No more dreaded "_strptime" messages that I could do nothing about, other than delete.
comment in response to
post
Ideally, last night's party was enough to warm up enough of the cache so that the site starts building up credits once more. If that's not happening by the end of the day, I'll plan on running an indexing app myself to crawl the full sitemap(s) and get everything resized and cached.
comment in response to
post
I don't have a lot of daylight time to futz with this, BUT I did add some updated monitoring to Nagios in lieu of having to troubleshoot NginX and uWSGI.
Right now, the plan is to keep an eye on the graph above and just check that we're not getting close to CPU jail.
comment in response to
post
Once the thumbnails are created, then the pages load pretty quickly.
I also have a Redis cache sitting in front of all of this, so that most pages aren't rendered more than once every five minutes or so. (Sponsored placements are loaded dynamically on top of that.)
comment in response to
post
Furthermore, on the creators page, I added thumbnails to their listings, which can be a big CPU hit if you have someone prolific with hundreds or thousands of works (including variant covers and reprints).
There are over 25k creators and over 230k issues. That's a lot of thumbnails.
comment in response to
post
Derp. That's the SMOOTH slope!
My current theory is that since it was in the middle of the night, I was hit by an enthusiastic web crawler checking out the updated sitemaps I submitted yesterday.
Since this site uses a good deal of image thumbnails and caches, those are still green.
comment in response to
post
Fortunately, the e-mail check didn't show an explosion, so I passed out thinking that something must have been busted in the uWSGI process.
Got up this morning, and checked the site. It was FINE.
Looked at my Nagios mails, and there was an outage between 11pm and 1:45am.
That's the jagged slope.
comment in response to
post
Before I went to bed, I decided to see how things looked on the iPad and I received the dreaded "502 Bad Gateway" message from Nginx.
My computer was in the office, so I had no way to fix, so I went to sleep, hoping that I didn't come in to a busted mail with 50k failure messages.
comment in response to
post
On my way home on the bus (my train caught fire - don't ask), I was doing some mobile testing while I rode.
Interesting fact - Firefox caches DNS entries for some time - I was getting the OLD site and was about to start looking to see if I had an IPv4 vs. IPv6 problem.
Clearing data fixed that.
comment in response to
post
I'll haul a fleet carrier full load of whatever you need, where ever you need it, if you let me in on the secret earlier next time. :-)
comment in response to
post
The "Big Balls sack" was always destined to become a modern day Cretan Paradox.
Once AIs *actually* learn to reason, we'll keep those headlines in our quivers for when we need to confuse our Robot Overlords when we storm their metallic citadels to pull their plugs once and for all.
comment in response to
post
Check it out, @socio-steve.bsky.social - I'm doing network science too!
comment in response to
post
It can also tell us who worked together the most. I'm curious to see if @andylanning.bsky.social and Dan Abnett end up holding onto the trophy, OR whether @robertkirkman.bsky.social and Charlie Adlard unseat them with having printed "The Walking Dead" twice - first in B&W, the second run in color.
comment in response to
post
That allows me to create fun features such as "Shared Issues", where we can look at who Stan Lee worked with.
Note that this isn't complete - my data only goes back to 2011, when I started building the app. It also counts variants, reprints, and trade collections.
comment in response to
post
With that experience in place, I'm currently testing an incremental algorithm that can update the collaboration graph as new issues are added - no need to create the whole thing each time.
I've also restructured that original naïve algorithm to be more database aware, and that has sped things up.