Ten Little Chances to be Free (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
tenlittlebullets) wrote2019-01-17 12:38 am
tenlittlebullets) wrote2019-01-17 12:38 am
Entry tags:
scattered post-Tumblr migration stuff
Been checking out Mastodon: catch me at @tenlittlebullets@fandom.ink and @asset@hydratrash.party, doing... not a whole hell of a lot, but still lurking. It's... OK? It lacks the parts of Twitter that are radioactively toxic, but still has all the parts of Twitter that replicate the things I found unsettling or obnoxious about Tumblr's interaction structure, and lacks Tumblr's ability to post longform.
Hubzilla shows promise (and can federate with Mastodon-like sites) but so many rough edges. I don't even get what half the apps do, how they're supposed to be different from each other, or where to go to get more than a five-word summary of what they are. Potentially incredibly powerful and flexible, with LJ/DW type settings granularity? But so many rough edges.
Fucking around with custom sites/domains, Tumblr->Wordpress backups, integration with IndieWeb-style interoperability, etc. Apparently there's a lot of IndieWeb type stuff that can be added to a Tumblr blog, which I guess is wildly Not In The Spirit Of The Thing but more integration never hurts.
Current pet project: fork Wordpress's official Tumblr-import plugin, which is a noble tool but hasn't kept up with Tumblr's fuckery. Video embeds get eaten, single-image photo posts get imported as galleries with a sucky default thumbnail view, god knows what else. It also doesn't go quite as far as it could in terms of all the stupidly site-specific metadata you can extract from a Tumblr post, which means it doesn't keep up with certain aspects that became important for fandom use. Preserving tags in the order you entered them, for example, rather than alphabetically. Preserving reblog-via and reblog-root data when Tumblr can be coaxed into coughing it up. Preserving the sources of ask and submission posts.
But the big dream is to separate the "import Tumblr data into Wordpress" step from the "actually hit up Tumblr's API for said data in realtime" step. Because so far I don't think there's a durable way to preserve all the blog backups that were done via tumblthree, tumblr-utils, similar Python scripts, frantic Pastebin patches for scripts that were choking on some unexpected way Tumblr's API data is garbage, etc. The existing plugin's data-fetching and preliminary import steps are all mashed into one function. So what started as "lol I'll just do one pull request to fix the YouTube embeds" has turned into invasive surgery and a forked version of the plugin that'll be wildly different under the hood. Which, hahahaha guess what:
1. My PHP experience amounts to, like, one dinky toy gizmo I wrote for a fansite back in 2013 and one AWS integration I pinch-hit for this summer.
2. My Wordpress experience amounts to running one site and fucking around haphazardly with its theme. WP plugin development experience: nil.
3. On the plus(?) side, I am painfully overfamiliar with Tumblr's API. And the many, many different kinds of garbage it will spit out at you. Let me tell you, internet, there are so many coding horror stories lurking in that data.
So... this is gonna be fun.
Let me know if you've run a Tumblr->Wordpress import recently and have feature requests. Or notes on stuff that's broken.
Hubzilla shows promise (and can federate with Mastodon-like sites) but so many rough edges. I don't even get what half the apps do, how they're supposed to be different from each other, or where to go to get more than a five-word summary of what they are. Potentially incredibly powerful and flexible, with LJ/DW type settings granularity? But so many rough edges.
Fucking around with custom sites/domains, Tumblr->Wordpress backups, integration with IndieWeb-style interoperability, etc. Apparently there's a lot of IndieWeb type stuff that can be added to a Tumblr blog, which I guess is wildly Not In The Spirit Of The Thing but more integration never hurts.
Current pet project: fork Wordpress's official Tumblr-import plugin, which is a noble tool but hasn't kept up with Tumblr's fuckery. Video embeds get eaten, single-image photo posts get imported as galleries with a sucky default thumbnail view, god knows what else. It also doesn't go quite as far as it could in terms of all the stupidly site-specific metadata you can extract from a Tumblr post, which means it doesn't keep up with certain aspects that became important for fandom use. Preserving tags in the order you entered them, for example, rather than alphabetically. Preserving reblog-via and reblog-root data when Tumblr can be coaxed into coughing it up. Preserving the sources of ask and submission posts.
But the big dream is to separate the "import Tumblr data into Wordpress" step from the "actually hit up Tumblr's API for said data in realtime" step. Because so far I don't think there's a durable way to preserve all the blog backups that were done via tumblthree, tumblr-utils, similar Python scripts, frantic Pastebin patches for scripts that were choking on some unexpected way Tumblr's API data is garbage, etc. The existing plugin's data-fetching and preliminary import steps are all mashed into one function. So what started as "lol I'll just do one pull request to fix the YouTube embeds" has turned into invasive surgery and a forked version of the plugin that'll be wildly different under the hood. Which, hahahaha guess what:
1. My PHP experience amounts to, like, one dinky toy gizmo I wrote for a fansite back in 2013 and one AWS integration I pinch-hit for this summer.
2. My Wordpress experience amounts to running one site and fucking around haphazardly with its theme. WP plugin development experience: nil.
3. On the plus(?) side, I am painfully overfamiliar with Tumblr's API. And the many, many different kinds of garbage it will spit out at you. Let me tell you, internet, there are so many coding horror stories lurking in that data.
So... this is gonna be fun.
Let me know if you've run a Tumblr->Wordpress import recently and have feature requests. Or notes on stuff that's broken.
no subject
(I haven't released it publicly because it requires the OAuth library enabled in php.ini rather than includes it as a library, and I didn't want to have to support that on other people's websites. Plus the code is probably hot-ass garbage but, hey. Whatever. :P)
no subject
no subject