-
-
Notifications
You must be signed in to change notification settings - Fork 223
-
Just hit 1 million!
(The number might dip below the mark again during the holidays though.)
I don't know whether there is a good tool to monitor a version number trends.
The best I can do currently is to compare my packages usage, since at the moment they are used almost exclusively as html-to-text dependencies.
parseley and selderee were introduced as the dependencies in the version 8.0.0 (2021年06月09日).
Looks like it took almost a year for it to occupy 50% of all downloads.
Trends are almost the same as html-to-text, just 400k downloads below.
I interpret this as there are 400k (currently 40%) of client systems that are actively used but unmaintained.
Or maybe some of them were held back by certain issues with version 8.x - we'll see how the trend will evolve.
Considering how the ratio of old versions changes during holydays and weekends (on the daily view), some assumptions can probably be made about what kind of businesses and institutions are more likely to run their systems without maintenance.
I switched parseley to my own lexer and parser combinators packages (leac and peberminta) and this change is shipped with html-to-text version 9.0.0 (2022年12月03日).
I can see that about 10% of client systems are updated very quickly.
Time to put weekly npm download number badges on my peberminta and leac packages - now they've been skyrocketed into some very credible numbers :)
@html-to/text-cli is now a separate package and I can finally see how actively the CLI is actually used.
Looks something like 0.2% of updated users. Not a huge amount, but still better than I anticipated.
Beta Was this translation helpful? Give feedback.
All reactions
Replies: 2 comments
-
One year later:
Observations
html-to-text version 9.0.x now represents the majority of downloads.
peberminta and leac are slowly diverging, the difference is 50k weekly downloads now. Considering the fact they are used together in my projects, it means there are pretty big users of peberminta now that I'm not aware of. There are no public NPM packages, meaning some "end of the dependency chain" applications use it. Still, pretty cool, and I'm most proud of it.
Looks like in February some big downstream project using old version of html-to-text experienced exploding growth.
Then, in August some big downstream project updated to version 9.0.x. This might also correspond with a small surge of issue reports when users are trying to configure nodemailer, mailparser or something like that by outdated instructions from the web and don't understand why it fails...
Should I consider those reports as a lesson to take backwards compatibility more seriously? Maybe, but also not really. There are enough changes between versions where backwards compatibility for the method in question existed to warrant a good look when migrating. If anything, long backwards compatibility cycle can actually hurt - people configure a new system based on old instructions and for them it is not even obvious why things break not immediately after using bad instruction but after time when updating my package.
Progress
All I did this year - published a couple of patch versions.
I'm still struggling with my life challenges (#284). Learned I have AuDHD amidst the situation that leaves me no solutions to deal with it.
Can't find the peace of mind required to work on my personal projects.
Reported issues are slowly accumulating. I hope I can resume some progress, but I don't know how realistic it is.
For a while I was out of sponsorship options.
Recently I was able to activate https://github.com/sponsors/KillyMXI - very surprising it was GitHub of all things - none of those coffee donation sites offer cash-out options I have access to.
(Still can't find time to update sponsorship links - I don't want it to be the sole entry in a changelog, such thing leaves a bad taste when I see it elsewhere.)
But with the kind of projects I have - it is unrealistic to have sustainable donations stream.
For now, the best way to really support my projects and get things done - still is to hire me.
Maybe someday I'll find something with wider support, but for now that idea still shatters over "no peace of mind" issue.
Beta Was this translation helpful? Give feedback.
All reactions
-
One year later:
image
Observations
Reached 2M weekly downloads 🎉
Absolute gap between versions (indirectly observed through upstream packages) is not shrinking - meaning old systems are not upgrading.
Relative percentage of old versions is dropping due to growth of new users.
Log scale graph shows that the usage of CLI is stable at around 1000 weekly downloads.
Graph is pretty smooth and doesn't have any more features worthy of note.
Progress
I'm still struggling with my life - some things getting mildly better, other getting worse.
I was unable to complete any updates this year. Looks like I'll be jumping many Node versions when I finally get to it.
I was hoping to publish new versions of leac and peberminta before New Year at least, but now this is unlikely to happen. leac is essentially ready, with some performance improvements, but since it is likely to be used together with peberminta, I wanted to publish them simultaneously. And there is a lot of chore remaining to be done for peberminta.
There is some case of software rot issue. Node users may experience some compatibility issues between some packages and modern Node versions. This situation bothers me a lot, but without any support I have no ETA to update everything.
Beta Was this translation helpful? Give feedback.