Jekyll Indieweb

My API key got revoked. I could sign up for a new one,
but Twitter is pretty dead now. I can't yet be bothered
with Mastodon. check-in: 08440d766f user: atomicules tags: trunk

Website is sizeable enough that simply wiping the destination and re-copying
files is a bit dumb. Use rsync and only re-create the compressed files (no way
around that); Although I have realised I could have been using the "keep"
option on gzip all along. Doh! check-in: eb930e542e user: simon tags: trunk

Doh! Need to get datetime to 00:00 as otherwise can still fail. check-in: 1ca70f9df1 user: atomicules tags: trunk

I never really bothered with future posts before, usually leaving things in a
draft folder instead. check-in: 7492f6da82 user: atomicules tags: trunk

Seems to work ok and does speed things up. Also add commented-out versions of
build and serve for when using git version of Jekyll. Add a comment about date
and time formats, must have changed at some point. check-in: 083b073943 user: atomicules tags: trunk

I.e. basically look for a yaml key of that name. No need for that key to have a
value. Just check for it existing as by default it won't. Adding this as some
photos/images I don't want to syndicate to Flickr. check-in: 75924f9e07 user: atomicules tags: trunk

So I can do strings like so: "And so I said, "I didn't do that!""

I have to remember to do this, but this is good enough:

rake note["And so I said, \"I didn't do that!\""]

References 3ffea62560f305c2c5cf188c146ed6f1271d62f6 check-in: 6de2cd4592 user: atomicules tags: trunk

Goes hand in hand with Jekyll include changes I've made for photo posts where
I'm now putting leading slashes for my image paths. Also change logic ordering
a bit to make more sense. check-in: 29b5f69899 user: atomicules tags: trunk

Hadn't thought that the Rakefile was getting copied to _sites and therefore
deployed to the web server.

Also, since even my basic web server supports serving compressed content I'm
now creating compressed copies of everything during deployment. check-in: e56656e837 user: atomicules tags: trunk

- Major refactoring of code. Previously each syndication class looped through
all the posts using find_posts_since, but it makes much more sense to loop
through the code once, directly within the rakefile and call the relevant
syndication objects within that.
- Require module within rakefile and require individual syndication classes
from module. This is a cleaner approach. Previously each syndication class
file required and included the module and the rakefile required all the
individual class files - crazy!
- Finally, use a hash to pass around all the post_data info as can just pass a
hash as an argument as opposed to passing individual arguments and forgetting
which order they are in and that some syndication classes required more
arguments than others, etc. This was because find_posts_since had to generate
the data correctly depending on what post_type it was being called for. With
a hash it doesn't matter if I pass additional unnecessary arguments as long
as I use the same key names everywhere. check-in: 4e162f5e7e user: atomicules tags: trunk

REALLY need to tidy things up now. check-in: b57a788793 user: atomicules tags: trunk

So only on link-type posts that are marked up as being u-in-reply-to.

In implementing this I've realised I really need to rethink how my code works.
It is now a mess. Each synidcation or webmention task should not iterate
through all the posts. That only needs to happen once and from that call the
appropriate syndication task, etc. check-in: 2bfd536be0 user: atomicules tags: trunk

Need to re-enable build tasks as I may need to re-build more than once during deploy due to adding in rel syndication links and I didn't realised Rake would only run them once.

Also, handy to report out number of posts syndicated under each type (photo, link, note) check-in: 34bcb284b2 user: atomicules tags: trunk

Use DateTime instead of Date, reorder some of the rake tasks, add a test task for testing notes check-in: f9ea46b2de user: atomicules tags: trunk

This is still all working on the assumption of one tweet per day

Haven't yet linked in the build, deploy and syndication bits. check-in: 0bbb623e09 user: atomicules tags: trunk

- Required editing the flickr syndication file to breakout the add_link_to_post method since that can be used for notes as well.
- kept the flickr specific bits in the flickr syndication file
- Made the method more flexible and added that to jekyll_syndicate.rb.
- Add support into find_posts_since for finding posts from the notes sub-directory. check-in: 907c8224e6 user: atomicules tags: trunk

I'm sure there is a neater way to do this, but for now making sure the Pinboard
instance I create is a class variable ensures it works everywhere. Weirdly,
this isn't necessary with the Flickraw gem, but I think that uses some kind of
voodoo. check-in: e9fa1d4d8e user: atomicules tags: trunk

The reason for doing this is so the NoPostsFoundError can be utilised. When
called as external ruby scripts it is basically a system call so you can only
see pass/fail. I want to have NoPostsFoundError as a fail, but a legitimate
one, so if it occurs it won't try to rebuild for the flickr syndication because
there is obviously no need, but also won't cause the whole deploy script to
fail. Previously the only way I could stop this legitimate error from stopping
the whole deploy was by rescuing any type of error. And that's not such a good
idea.

In order to get this to work I had to turn everything in the flickr and
pinboard syndication files into classes instead of having them suitable for
command line scripting. The Rakefile then needs to require this and create
instances of these classes.

As a result of notes support that I'm mid implementing on my site I found an
error in `find_posts_since` as I was assuming everything under _posts was a
legitimate file whereas potentially I could now have sub-directories which
obviously don't date parse. check-in: 98c0599f8f user: atomicules tags: trunk

I.e. anything pushed out to flickr is then automatically edited to add in the
link to the flickr url. At the moment this url is the nsid one rather than the
username version. Need to work on this. There's probably other bits that need
doing as well, but I've been drinking and that isn't such an often thing for me
so... well, I have tested this code, but not sober. check-in: d1a4052969 user: atomicules tags: trunk

Although I have achieved my goal of DRY-ing up these two methods I don't
necessarily think I've made nice code. For instance, if I decided to add
another one of the Pinboard API methods in, would this extend to that nicely?
I'm not sure.

However, since this code is just for me I get to play a bit.

I am glad I've managed to avoid eval and eveything is "proper" code though. check-in: b1f6733590 user: atomicules tags: trunk

See [7128bce9c0] check-in: 93a0970f92 user: atomicules tags: trunk

So I can have consistency in module and filenames, etc check-in: 4020a4bbaa user: atomicules tags: trunk

Only roughly tested. Should be mostly ok though. check-in: a4a9440169 user: atomicules tags: trunk

Uses Flickraw since it is such a good gem. I didn't use anyone elses gem for
Pinboard because the last time I looked none of them supported the required
rate limiting.

Tested it and it does work, but lots still to do:

- Move out code common to Pinboard and Flickr to another file
- Search flickr first before posting for the POSSE machine tag
- Update the blog post before deploying with the photo ID from Flickr check-in: 2925e8eba4 user: atomicules tags: trunk

This was in another repository, but I might as well move here since it ties in
well with the Indieweb stuff. check-in: bfdf6ceae9 user: atomicules tags: trunk

Was meant to be this way anyway. Commit previous value by mistake. check-in: d3e8c0a35a user: atomicules tags: trunk

- Use Logger class instead of puts or returning messages in array
- Correct arguments for replace parameter in Pinboard Add method. Should be
"yes" or "no", not "true" or "false"
- Correct statement about minimum requirements for Pinboard Add method. Need
url and title/description
- Make Pinboard Add method have all arguments optional apart from url and
title/description check-in: 7128bce9c0 user: atomicules tags: trunk

Also do the array of tasks in a more Ruby-ish way. check-in: 5817b97d07 user: atomicules tags: trunk

Exec doesn't return, hence why my deploy task was only running the first of the
dependencies. I then read that dependencies aren't executed in any specific
order so decided not to use them and just invoke them all instead. check-in: 37ebd77332 user: atomicules tags: trunk

- Add option parsing, etc so I can call this script from the command line
- The posse tag wasn't being added for new posts due to my own insanity it seems. Must have been confused. check-in: bbce182326 user: atomicules tags: trunk

It works!

Need the corresponding deploy script. Also still tidying up to do. check-in: 59b46a58a6 user: atomicules tags: trunk