Kept the old hacks where they don't break the build in case they things
they fix are still relevant.
I checked that the upgrade doesn't break:
1) Asymptote and EProver builds.
2) My XeLaTeX demo from configurations/ repository.
3) Some of my own files.
The upgrade fixes problems with simultaneous use of 3D and LaTeX labels
in Asymptote.
Please provide a test that worked previously and is broken now if you
need to revert this update or its parts.
The current asciidoc expression is impure; it relies on several tools to
be found in PATH at runtime. This commit adds a enableStandardFeatures
parameter that, if true, pulls in all dependencies and patches asciidoc
to contain full paths to the tools.
I've set enableStandardFeatures = false for the existing asciidoc
attribute so that the closure size stays unchanged, at 255 MiB. The new
asciidocFull attribute (with enableStandardFeatures = true) has a
closure size of 1.5 GiB.
imagemagick, transfig, inkscape, fontconfig and ghostscript are missing
dependencies of dblatex. Instead of adding all those dependencies to the
existing dblatex attribute, make a new dblatexFull attribute for that.
Also pass --use-python-path at install time so that script shebangs end
up with #!/path/to/python instead of #!/path/to/env python (which is
impure when not run in a wrapper).
* Remove package name
* Start with upper case letter
* Remove trailing period
Also reword some descriptions and move some long descriptions to
longDescription.
I'm not touching generated packages.
There are many more packages to fix, this is just a start.
Rules:
* Don't repeat the package name (not always that easy...)
* Start with capital letter
* Don't end with full stop
* Don't start with "The ..." or "A ..."
I've also added descriptions to some packages and rewritten others.
Rename NetworkManager-dispatcher.service to
dbus-org.freedesktop.nm-dispatcher.service so it is found by systemd (NixOS
doesn't handle the Alias directive in the unit file)
This reverts commit 0350bd3b48. It
causes a huge increase in the closure size of dblatex, since it now
depends on GUI packages like Inkscape. Also, statically depending on
teTeX might be annoying for people who use TeXlive.
This reverts commit aef81d6eb6.
It's really not good to have every little package that depends on
asciidoc to pull in 1.5 GiB in dependencies (such as Lilypond).
The current asciidoc expression is impure; it relies on several tools to
be found in PATH at runtime. This commit adds a enableStandardFeatures
parameter that pulls in all dependencies and patches asciidoc to contain
full paths to the tools.
enableStandardFeatures defaults to true because asciidoc may attempt to
call all tools in its default configuration. With all standard features,
the closure size increases from 255 MiB to 1.5 GiB. Set
enableStandardFeatures = false if you want a minimal asciidoc.
imagemagick, transfig, inkscape, fontconfig and ghostscript was missing.
And pass --use-python-path at install time so that script shebangs end
up with #!/path/to/python instead of #!/path/to/env python.
pass is just a simple script to store passwords encrypted via gnupg in
a git repository. It uses many other tools, and until now relied on
them being in PATH.
This commit wraps the script and sets PATH.
See https://github.com/dagwieers/asciidoc-odf for more info.
This plugin has a shebang with /bin/env instead of /usr/bin/env, so update the
patchPhase to handle both cases.
WARNING/TODO: Libreoffice says "General Error. General input/output error."
when I try to open the generated .fodp files. So the odt backend works fine,
but the odp backend does not.
Also, slightly change the shebang fixup in the patchPhase so that it
handles optional [[:space:]] before the interpreter path (needed for the
filters).
To enable the extra filters, put this in packageOverrides:
asciidoc = pkgs.asciidoc.override {
enableDitaaFilter = true;
enableMscgenFilter = true;
enableDiagFilter = true;
};
The previous version (5.0.26) has been removed from the debian ftp.
As the source URL is now down, our own hydra (not hydra.nixos.org) failed
to build the package.
This problem will occur again in the future since I only updated the URL
without relying on a "more stable" alternative (this merits a specific
discussion).
Also some style cleanup.
Note that defining an empty-string variable *does* change the hash.
I would like to change this behaviour one day
(clean up attrs when compiling the derivation).
From https://lists.macosforge.org/pipermail/macports-dev/2011-July/015263.html:
5) Building with a compiler that doesn't support newer __builtins
If your port uses MacPorts compilers rather than the default compiler,
you may run into trouble with string functions. You'll see errors at
link time about undefined __builtin_* functions. If this happens, you
may want to compile with -D_FORTIFY_SOURCE=0 to tell the headers to
use unfortified versions which do not use compiler builtins.