Service Update

Author: Ben <benk@tilde.team>

Mon 19 Dec 2022 11:15:13 AM +05

This post is in part to test some tweaks that I've recently made to my setup. I finally got fed up with gemfeed's misbehavior when it comes to improperly sorting gemlog entries in the resulting atom feed. A number of times changing computers has led to the order being broken and throwing aggregators like Antenna through a loop. Therefore, I am putting a hold on atom feed generation for now and reverting back to submitting the gemlog as it is in gmisub format. The reason I used atom at all was just to make sure that my posts were sorted properly in Antenna if I wrote too many too fast, but let's be realistic... how often am I going to write more than one post in a day? It happens so infrequently that it doesn't merit relying on broken software to try to give it unnecessary finesse.

Another reason to use atom is for Capcom, but it's been a while since I've checked that site, and I don't consider it vital, although it's a great aggregator. Atom will be back sometime in the future, but I would like to write my own generator that works reliably with my gemlog format. (Any program could generate an atom feed of the files as they are since their filenames include the date and time they were published already.)

Earlier in my gemlog I wrote about how I couldn't figure out why "echo -e" was echoing "-e" into my gemlog posts on my Chromebook (Debian Bullseye). It truly stumped me at the time, but I randomly discovered while Googling that some shells use their own built in "echo" rather than the echo command provided by the system, and I guessed that running the script with /bin/sh on Debian was causing it to run in some sh compatibility mode. Just changing the shell executable to bash instead of sh solved the problem completely. Very sneaky, GNU.

The last thing is that I recently switched my Gemini server from vger to gmid. The reason for this was that after vger was upgraded to 2.0.0 on FreeBSD, it stopped working on my machine as it was supposed to. (The chroot function was broken.) It is only one of a few problems vger exhibits on FreeBSD, and it turns out that gmid just does the (same) job more reliably. Maybe vger is technically more secure (maybe not), but gmid is really a lot more convenient and comaptible on this sytem. I may consider switching back to vger if the issues with FreeBSD get solved, which I'm currently in correspondence with the author to address, but I have to say gmid has made me very happy in the mean time!