| Commit message (Collapse) | Author | Age | Lines |
| |
|
| |
|
| |
|
|
|
|
|
| |
Sadly we'll have to move around the footnotes here, but the post is
young enough for that to hopefully not be an issue.
|
|
|
|
|
|
|
|
|
| |
When generating archives, git-archive(1) puts the commit ID in the
global extended pax header. Therefore, an interesting side effect of
signing tarballs generated with git-archive(1) is that we also
implicitly sign a commit. Whilst we don't expect anyone to go through
the hurdles needed to verify such a commit signature, make sure to
document this case anyway.
|
|
|
|
|
|
|
|
|
| |
freenode is no more [1][2][3][4].
[1] https://libera.chat/news/welcome-to-libera-chat
[2] https://fuchsnet.ch/privat/fn-resign-letter.txt
[3] https://www.kline.sh/
[4] https://mniip.com/freenode.txt
|
|
|
|
|
| |
Since the previous commit already uses examples from this repository, go
all the way and use it for the rest of the post too.
|
|
|
|
|
|
|
|
|
|
|
| |
This commit attempts to clarify what exactly a tree is, and how trees
used for notes relate to "regular" trees. To that end, include real-life
examples and output from commands run in a repository that is publicly
accessible (in this case, quarg). This allows people who are interested
in the internals to follow along and poke around in the repository
themselves.
Many thanks to V for initial feedback and motivation for this change.
|
|
|
|
|
|
|
|
| |
These were here because the argument used to contain a glob, like so:
--notes='signatures/*'
Thanks to V for reporting this.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit 751da2c (posts: Update posts to reflect usage of cgit's signature
mechanism, 2021-05-06) updated the project pages as well as "Verifying
snapshots with signify" with information regarding the new cgit
signature mechanism, but neglected to mention that signatures can be
obtained directly through git.oriole.systems as well (the project pages
merely deeplink to that).
Furthermore, with commit 28c2473 (posts: Add a new post: "Hosting
signify signatures on cgit", 2021-05-16) we now have an explanatory
piece which we can link if people are interested in the underlying
infrastructure.
|
|
|
|
|
|
| |
Until now we were using a system-wide httpd, making changes to its
configuration file a bit onerous. Instead include a small Caddyfile we
can use together with a call to 'caddy run' for local development.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Let's hope that bug is fixed in the next 10 years!
|
|
|
|
|
|
| |
The -p flag didn't really make a lot of sense unless entr runs the
entire time. Additionally, sleep for a second in the loop to make it
easier to cancel.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
cgit can host detached signatures using git-notes(1). Since we want to
make use of that going forward and have finally gotten around to setting
it up fully, we need to make some small changes to
verify-with-signify.md and the weltschmerz project page.
Specifically, we no longer host both the signature and the checksum
(since this would be too cumbersome to maintain with cgit), rendering
the choice explained in the post moot. Since the detached signature has
always contained the checksum, however, we can just fall back to a
slightly different invocation for verifying it.
|
| |
|
|
|
|
|
|
| |
'align-self' cannot operate on elements that are not flex. This is most
probably a remnant of an earlier discarded design whilst implementing
77bc0a9 (Attempt a thorough redesign of the landing page, 2020-06-06).
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In commit bbaaf25 (templates: Add date and author to head metadata,
2020-04-10), a <meta> tag containing the publication date was added for
unknown reasons. It is unclear where this line is from, as the 'scheme'
attribute is not even part of the HTML5 specification. Furthermore, the
value of "date" for the 'name' attribute is not standardized.
Use "dc:date" as per the Dublin Core Metadata Initiative [1] instead of
"date" and remove the 'scheme' attribute fully.
[1] https://www.dublincore.org/specifications/dublin-core/dcmi-terms/#http%3a%2f%2fpurl.org%2fdc%2fterms%2fdate
|
|
|
|
|
|
|
| |
This should make it easier for screen readers to build a page outline
and is recommended [1] by W3C.
[1] https://www.w3.org/wiki/HTML/Usage/Headings/Missing
|
| |
|
|
|
|
|
|
|
| |
Caddy supports implied file extensions in 1.0 via the 'ext' directive,
and in 2.0 via the 'try_files' directive. This feature has been enabled
in the upstream web server for quite a while now, but the links were
never updated to be cleaner.
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Add support for prefers-color-scheme to man.css by including all custom
properties and referencing them in man.scss
|
|
|
|
|
| |
In preparation for making man.css respect prefers-color-scheme, move all
custom properties to their own file so they can be included easily.
|
|
|
|
|
|
| |
These were skipped because the round borders would not be visible on a
white background. Now - with a dark colour scheme - they are, so make
them round.
|
| |
|
|
|
|
|
|
| |
This element's foreground colour is inherited from the root element.
Since we use a special background colour here, make sure to explicitly
set the foreground colour in conjunction.
|
|
|
|
|
|
| |
Currently we use the default colour for these. In preparation for
prefers-color-scheme, add two variables containing the default colours
for normal and visited anchors.
|
|
|
|
|
|
| |
Currently, colours are set per element. In preparation of the support of
prefers-color-scheme, use variables instead so they can be modified in
one place.
|
|
|
|
|
| |
This particular background colour will be inherited from the root
element and does not need to be defined explicitly.
|
|
|
|
|
|
|
|
| |
In preparation for the support of prefers-color-scheme, convert all
presently used Sass variables to CSS custom properties. This will enable
us in an upcoming commit to define different values per media query.
While at it, use a better name for variables holding colours.
|
| |
|
|
|
|
| |
May the Oriole forever grace our skies.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
After careful consideration we have decided to replace the more
brutalist landing page with something simpler and more in line with the
styling of the actual posts.
Have the front page include a short introduction and some further links
to the atom feed and git.oriole.systems. Use a more traditional and
simple top-to-bottom layout for the article links in lieu of the more
idiosyncratic left-to-right list that was difficult to scan for some
people.
Keep the post styling as is, but move around and change a few HTML
elements to make more semantic sense and improve the clarity of the CSS.
|
|
|
|
|
| |
Introduce a small script that uses entr(1) to rebuild the site in the
every time a file is changed.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
When clicking on a footnote reference, the browser scrolls all the way
down to the footnotes block of the post. It is not immediately clear
which footnote was referenced, so use the `:target` selector in the
footnotes block to make the linked footnote bold.
|
|
|
|
|
|
|
| |
In fde2230 (css: Add styling for footnotes) we introduced styling for
footnote references, removing the underline and making them bold.
However, as pointed out by a user, having them bold makes them less
readable and clear. Therefore, reduce the font weight back to normal.
|
| |
|