| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
docs/iconcache.txt says the format is machine-independent.
It does not seem to result in any noticeable improvement.
|
|
|
|
|
|
|
|
|
|
| |
- Install a link to sdgui to the Start Menu.
- Associate the .ifo file extension with sdgui.
- Change the installation directory name to contain a space,
rather than a dash, since that appears to be the norm.
It's also copied over to the display name.
And thus, the GUI has become somewhat usable on Windows.
|
|
|
|
| |
At minimum it documents the novelty of the command argument.
|
| |
|
| |
|
| |
|
|
|
|
| |
CMake does know how to be tough on users, but we've won at last.
|
|
|
|
| |
But only if the respective binaries can be found.
|
|
|
|
|
|
|
| |
Remove most unused Adwaita icons and cursors, it's about 30 mebibytes
of raw bullshit that doesn't seem to be missing in any shape or form.
Rather sadly, the script is too complex to be rewritten in CMake script.
|
|
|
|
| |
Added a match against the version stated in the README.
|
|
|
|
|
|
| |
This saves 21 MiB of MSYS2 libraries, without any adverse effects.
The MSYS2 build remains bloated, due to the Adwaita icon theme.
|
| |
|
|
|
|
|
|
| |
No one bothered to ask whether it /should/ be done.
The hamburger needs to be replaced with a file open dialog there.
|
| |
|
|
|
|
| |
Neither StarDict nor shared-mime-info have their own MIME DB file.
|
|
|
|
| |
It's finally not horrible.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Nothing in GTK+ appears to be suited for what are virtually infinite
lists. Our workaround with GtkLabel and GtkScrolledWindow has been
heavily suboptimal and needs to be replaced.
Use Pango directly to handle our relatively simple needs.
Upgrades:
- the widget can be scrolled,
- keywords are repeated for each definition line,
- definition lines are now wrapped, and support 'g' and 'x' fields.
Downgrades:
- text can no longer be selected, so far.
|
| |
|
| |
|
|
|
|
| |
The original one is a horrible thing. Now we're self-reliant.
|
|
|
|
|
| |
Now all error messages produced by tools should be in the right
encoding, even if the system isn't in UTF-8.
|
|
|
|
| |
Their usefulness was almost negative.
|
|
|
|
|
|
|
|
| |
Writing DocBook XML by hand is an awful experience and the tools aren't
much better. Asciidoctor does it well. There's no need to worry about
semantics, man(1) just needs to be able to show something at all.
This project's manpage is sadly almost useless right now.
|
|
|
|
| |
A nice, round number. This allows us to remove some boilerplate.
|
|
|
|
|
|
|
| |
It has a potential to stay simpler than the TUI,
while having a wider feature set.
Not building this toy by default, it needs some time investment.
|
| |
|
| |
|
| |
|
|
|
|
| |
Not very useful.
|
|
|
|
|
|
|
|
|
|
|
| |
As it happens, there is no real need to constantly poll for changes,
since XFixes can inform us of updates as they happen.
With GTK+ gone we've got dependencies and error handling under control.
XCB is a truly awful thing to learn, though.
Our method will never work on Wayland or Windows, so we don't miss out
on anything by abandoning the huge toolkit.
|
|
|
|
|
|
|
|
|
| |
I've come to the conclusion that copyright mostly just stands in the way
of software development. In my jurisdiction I cannot give up my own
copyright and 0BSD seems to be the closest thing to public domain.
The updated mail address, also used in my author/committer lines,
is shorter and looks nicer. People rarely interact anyway.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
A lot better than that StarDict shitfuckery.
|
|
|
|
| |
Depends on GTK+, which should make it work with all of X11, Wayland, and Mir.
|
|
|
|
| |
Apparently we do need to call the "outer-level" function.
|
| |
|
| |
|
|
|
|
| |
Also get rid of some silliness that I'm only able to see now.
|
|
|
|
|
|
| |
Now the tools don't get built on `make all' but instead have their
own target called `tools'. It might be reasonable to move them into
their own directory sometime, instead of cluttering `src'.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Provides pseudo-random access to dictionary files compressed using dictzip.
It doesn't implement a cache, it just loads missing chunks until it has the
whole file. I'm not sure if discarding not recently used chunks is really
a useful feature. If there _was_ a way to get noticed when system memory
is low, I think the best way to handle that event would be to simply release
it all.
All in all, this is pretty useless. But it was interesting to write.
This has yet to be integrated into the application proper.
|
| |
|
|
|
|
| |
It's not perfect but seems to work well enough.
|
| |
|