Note that at this time, it's backend is still considered experimental.
It also requires Google's shader compiler to build, which is available
from https://github.com/google/shaderc/blob/main/downloads.md. This is
also known as 'glslc'.
We use <application> in the markup for all package names. Do that
consistently for rendered text on these pages, except when referring
to a version of a package.
Italic is a style, not a face - make the fontconfg page match
the TTF-and-OTF-fonts page.
Do not use caps for 'variable', it is not a proper noun.
Typos.
Change one or two 'normal' to other words: too many uses of normal
throughout the page. Mozilla's CSS doc describes weight 400 as
'Normal (Regular)', so for weights normal is the correct word
(Regular only applies to non-book non-oblique non-slant styles).
Rework GNU Freefonts text, Noto 'Latin' fonts include all recent
latin and cyrillic glyphs, the Freefonts only have minority glyphs
which were in Unicode in 2012. Noto is now preferred, so can
provide thos uncommon glyphs if it has been installed.
tuning-fontconfig - link directly to the Samples PDFs of my website
for viewing the aliases, as was already done under the Liberation
fonts in the TTF-and-OTF page.
Remove infromation about hinting from TTF-and-OTF-fonts page,
adding a comment to clarify that infromation about fontconfig
is on the previous page (in case anyone decides to install a
font and lands on the TTF-and-~OTF page).
Explain about font weights, and how installing only one or two may
let you slightly darken (or lighten) the text.
Comment about static fonts in files which also include variable
fotns - variable fonts are mentioned on the TTF-and-OTF page but
the data fits better in tuning-fontconfig.
Details eventually found (from 2020), with a very ugly set of bug
responses.
Note: Putting the commented link to the bug on a new line adds a
blank line to the rendered file, and there is already a blank line.
Moving the comment to the same line as the text fixes that.
Many older fonts, and even some currently-developed fonts such as
Junicode, lack hints. Before exploring hinting options it makes
sens to check that the font being used does indeed have hints.
Placed in 'Useful Commands' ahead of the Pango example, because
it is a plain fontconfig command.
In talking about hinting, mention dots per inch - some people can
detect colour fringing if the actual DPI is a little different from
96 dpi, and general recommentations for High DPI screens are to
disable hinting because it is not required when the font is increased
in size to have the expected size, i.e. more pixels are used for
the glyph so they can be either off or on rather than shades of grey.
A lot of the information which shows up in google, particularly from
Arch users, is for using the Infinality True Type interpreter. Let
people read the history from FreeType.
Also confirm, from a posting this month, that medium hinting is broken.
As a consequence, simplify 'other non-latin alphabets' to
'other alphabets' rather than 'other non-Latin alphabets'.
Correct the link to my own 'Substitute latin fonts' item, which
remains lowercase, to go directly to it and therefore make a
separate link for the font pages of that site as a whole.
To stop people thinking that every website's choice of font can be
easily overridden (firefox can do that, not sure about other
browsers).
This is about halfway through the commits from my initial private
branch.
Explain what 'latin' means in this context.
Detail all the font types mapped there.
Mention that 49-sansserif is where an unrecognized font is assumed
to be Sans.
This module fails to build on i686 due to an issue with the bundled
version of the PhysX SDK.
Since we're already disabling Qt3D due to a problem with the bundled
copy of assimp, let's just skip the Qt Quick bindings as well.
This is a rough version of the new Qt6 page. The installation
instructions should be OK, but the Dependencies, Contents, and
Short Descriptions need to be checked.
The instructions have not yet been checked on a systemd system.
There is no ChangeLog entry yet. It will be added when the page
has been validated.
Fix up libadwaita to build with it. Update the command explanation to
allow building it with Qt-5. Also fix the errors detected in
org.linuxfromscratch.lfs.xml reported by "appstreamcli validate".
I had a setup like this on one of my machines, now that I'm
looking at the detail of fontconfig in a local branch I discovered
that there were certain problems with the example:
1. I'm in an en locale, for pages that do not specify a locale
(or in vim/view, e.g. in mutt) the Japanese fonts were being
preferred.
2. Fontconfig does not consider UMing suitable for zh-sc so it
was hardly ever used - and it does not really belong in local.conf.
3, Really prefer a Japanese font for Sans Serif and monospace, but
no point listing two of them.
3. Comment where WenQuanYi Zen Hei is regarded as adequate and
therefore do not include it in these preferences, since it will
be picked up after them.
libxkbcommon-1.6.0 removes some definitions that are unused.
These definitions are referenced in qtbase so we remove them
with a sed for both the full qt5 package and gt5-alternate.
neither firefox nor epiphany can download them, and they are not
well maintained, because rarely tested.
This is WIP because the "(HTTP)" part of "Download (HTTP)" will
need to be removed too.
But let's see what users think first...
It now unconditionally runs gtk-update-icon-cache (or
gtk4-update-icon-cache) after installing the icons. As a result, we need
something to install gtk-update-icon-cache.
For the benefit of jhalfs, I've set gtk3 as a normal dependency, but
gtk4 as a nodep :)
Thanks goes to Joe Locash for the report.
When the patch can be converted to a not-so-long sed, we prefer the sed
because it tells people "what this command is doing" more explicitly and
also reduces an additional download. And for patch or sed we need a
<para> describing "what it fixes".
It's also a bad idea to fold the patch/sed command just before "meson"
in the same <screen> because it'd be too easy to overread it.
It's optional for the packages that use it, and they only use it to
provide additional support for profiling.
Sysprof now requires two more packages which are specific to it, so
let's archive it.
- make change from /usr to opt/xorg "nodump"
- make the removal of "-nolisten tcp" "nodump"
- narrow the change from none to on to the "Numlock" line: otehrwise
the comments about this option become incomprehensible:
(e.g. "option can be on, off, or on")
If xmlto is not installed and -Dgtk_doc=false, there will be no
documentation installed.
Also add a rm -rf command so we won't do things wrong reinstalling the
package.
We recommend using the /usr prefix for xorg, but the instructins
for putting it in /opt/xorg do work. For jhalfs, having optional
instructions is confusing, so we mark the /opt/xorg "nodump" so
they will be ignored when automating BLFS.
I had the Intel Media Driver installed on this system at some point in
the distant past, and it installed /etc/profile.d/intel-media.sh - which
overrode the libva-intel-driver, and caused gstvaapi.so to fail to load
correctly on Wayland... which then caused GDM to fail to start. Note
that X11 worked fine because of using a different code path.
If we had the Xorg Drivers section still with xf86-video-nouveau there,
this probably would've gone over there originally, but this place fits
well.
The original plan was to put this into LFS, but I decided against it
since it works fine until you start using programs such as Epiphany or
KDE Plasma.
