Switch to using the full screenshots in the metainfo files. These URLs should not change.
I've checked them, and I've added a comment that should help with the confusion as a result of their being both PNGs and JPGs. My impression is these should be pulled infrequently such as when flatpak and Linux repo packaging is done each Freedoom release, or Linux distro release.
These should now be "forever" URLs linking to image files that can be overwritten with each release without touching these XML files each time.
**Linking to the thumbnails is intentional.** Unlike the full screenshots the shrunken jpegs will always be smaller than pngs for these, so we never need to change the file extension; and there's really no reason for making everyone download huge screenshots with this stuff anyway - they can go to the website for that.
The current website organizes everything so that we have 8 screenshots of each game, numbered 1-4 for Chocolate and 5-8 for GZ. Including screenshots 1, 2, 7 and 8 therefore should give us a fairly representative variety.
A generic term like this isn’t really suitable for trademark, nor
likely to ever be disputed. The community and the engine both heavily
bias toward “episode” and indeed most other games do.
Closes: #682
GNOME Shell normally doesn’t display enough of the name on-screen,
leaving “Freedoom: P…” on the icons and making it impossible to
distinguish the games by name.
While not entirely ideal, just having “Phase 1” and “Phase 2” should
be decent enough.
GZDoom gets bumped up to second for its support for high screen
resolutions. PrBoom-Plus gets bumped to last-place for its menu
complexity. Eternity got removed due to its difficulty of building
and running on Linux.
ASCIIDOC and ASCIIDOC_MAN have been added, which default to the
(Python-based) reference implementation, but may be changed to
asciidoctor simply.
The manpages have been modified to use single-line section headings
for AsciiDoctor compatibility.
Current Desktop Entry and AppStream standards recommend using
reverse-DNS for all relevant files under them.
Adds some more scaffolding targets for install/uninstall, but there
aren’t easy text transformations to the new names.
Bunch of logic to use XDG_DATA_HOME or fallback according to the
actual XDG recommendations.
Use an array to build up paths (can be edited easily in the future or
for whatever custom purpose), which shellcheck warned is a bash-ism,
so change the script interpreter to bash as well. Let’s hope that
doesn’t break anyone (does anyone lack bash even if they don’t use it
as their shell?)
Python 2 is very near end-of-life, and Python3-compatible changes to a
few scripts introduced compatibility problems with 2.7 again. It went
unnoticed for me since my system symlinks "python" to "python3", but
it broke the build on systems where that symlink is still python2. At
this point in time, I feel it is worth targetting modern Python and
forgetting about 2.7.
Engines such as Odamex may not have a default search path and simply
search for freedoom1.wad relative to the current directory. This can
cause an unexpected WAD file not found error when executing the freedoom
script.
This issue is mitigated by setting a default DOOMWADPATH environment
variable with sensible search paths, if it is not already set. All
engines listed in the PORTS variable support the DOOMWADPATH environment
variable, which makes it a simple and convenient way of helping prevent
this error.
Organize .gitignore by moving all patterns into a top level sorted
.gitignore file. With this change both "git status" and
"git-ls-ignore-index" should return cleanly. The later checks if any
files in the index are ignored.
Using the black code reformatter, pass it over all our Python files.
This allows for a consistent style across the code base.
Exception: lumps/dmxgus/stats.py, for readability.
PORT is too generic, especially if refering to TCP/UDP ports, and if
those are set, the freedoom script may not work properly. DOOMPORT is
a reasonable alternative.
Also includes a change to the top-level Makefile to accommodate new
target names. Moved the installation variable settings close to the
install targets.
Non-reference AsciiDoc implementations such as AsciiDoctor (used by
GitHub) or even the simplistic parser in loccount choke up on these
characters for different reasons. AsciiDoctor doesn’t seem to even
parse them, displaying them as-is and not with emphasized text.
loccount would get hung up on possessives, expecting a terminating
character for emphasis that never comes.
Annoying, but easy to work around. the curly ’ character can be used
for possessives and _underscores_ can be used for emphasis.
Given that there are a handful of animations that are fundamentally
incompatible between Doom II and Final Doom, and we have decided that
the correct thing to do is to prefer Doom II compatibility.
Additionally, there are only maybe a half dozen Final Doom mods around
and they pretty much all focus on being unofficial sequels. We won’t
remove nor continue to improve the existing TNT/Plutonia-specific
resources, let’s just make it a little less of an issue when those
mods are broken (assuming it’s due to the Doom II conflicts...).
‘boom’ and ‘doom’ represent Debian-style links for the alternatives
system to set up preferred ports. Now that we no longer require
Boom compatibility, it should be safe to remove the former. This
should only be an issue if chocolate-doom -> doom, and tries to play
one of the maps that aren’t vanilla-compatible yet.
zdoom and prboom are dead ports, superceded by gzdoom and prboom-plus.
I hope their removals aren’t controversial. If it adversally affects
anybody: setup the ~/.doomport symlink to point to them.
Finally, crispy-doom. Odamex is still preferred before it because of
its simpler multiplayer capabilities, otherwise Crispy would have been
a winner (imo). It has a close relationship to Chocolate Doom, but I’m
leaving that last because of the non-vanilla compatibility of Freedoom.
Chocolate Doom is a last effort to run something.
The tags are shorthand for the license of each file and avoid
copying the full license text into each one (and avoids having
to manually update the dates in each one...).
GitHub’s AsciiDoc parser barfs on the standard AsciiDoc syntax to
represent such characters with only ASCII… Well, everybody should
support UTF‐8 these days. Let’s just use these directly.
