| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
| |
Variables defined on the boot commandline have preference over
any variable set in the .cfg file.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
With the goal to make the 'min' module smaller so that potentially
we can compress it with zstd instead of xz and thus minimize the size
increase of the resulting squashfs module (~10%).
Quick benchmark with just the 'min' module zstd-compressed and the rest
xz-compressed shows more than 10% reduction in the boot-up time of the
XFCE Live OS. Unfortunately, this will add 20 MB to the ISO size which
will have to come from somewhere in order to keep the ISO size below
703 MB.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The new zstd compression support for squashfs modules allows for a much
snappier system because zstd decompression is 5 times faster than that of
xz (see https://sourceforge.net/p/squashfs/mailman/message/35989805/), but
at the expense of ISO sizes that are at least 10% bigger. The latter means
that we have to do some more pruning of the XFCE and PLASMA5 images to make
these fit in a CDROM and DVD image, respectively.
Here are some initial benchmarks with PLASMA5 Live ISO images,
booting in QEMU (times are in minutes:seconds):
Stage ZSTD XZ
---------------------------------------
Boot to runlevel 4 start 0:39 0:50
SDDM Login Manager visible 0:55 1:26
Plasma5 desktop ready 1:42 3:17
I also tested two of the bigger applications, measuring the time between
entering the command in a terminal and having a usable application window:
Stage ZSTD XZ
----------------------------------------
LO Writer window visible 0:17 0:34
Chromium window visible 0:09 0:14
It is obvious that a big Desktop Environment like Plasma5 where a lot of
binaries have to be loaded from their squashfs modules benefits a lot
from zstd, because after booting, the Plasma5 DE is available in
roughly 53% of the time it takes when using xz compression.
Zstd support in squashfs was added to Linux kernel 4.14. That means,
no customimzation of Slackware is required to make the Live OS work from
zstd-compressed squashfs modules.
In order to *create* these zstd-compressed squashfs modules, you will need
some custom packages at the moment, until they get added to Slackware-current:
zstd and a rebuild squashfs-tools to add the lacking zstd support to
mksqhashfs/unsquashfs.
|
| | |
|
| |
|
|
|
| |
The old use of 'installpkg' prevented upgrading an already present package
while 'upgradepkg --install-new' allows for all options.
|
| |
|
|
|
| |
Also, split 'x_base' into 'noxbase' and 'x_base' so that in future
we can create a 'rescue' option using just 'min' and 'noxbase'.
|
| |
|
|
|
|
|
| |
These wallpapers don't really add value to the ISO and when using the
new supported compressor 'zstd' we need to trim the ISO size down to
the max size of a DVD.
The zstd compression creates ISOs that are ~10% larger than when using xz.
|
| |
|
|
|
|
|
|
|
| |
This will provide compression ratios approaching those of xz,
while achieving decompression rations similar to lzo (5 times faster
than xz).
We will need to have a zstd package in Slackware first, and then
rebuild squashfs-tools to add zstd support.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
The elilo package in Slackware 14.2's ./patches has our liveslak patch
to eliloconfig already incorporated.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
If we use 'testing' repository to get the Plasma 5 packages, then the
slackpkg/slackpkg+ configuration should reference that 'testing' URL
as well, to avoid confusion when the user runs 'slackpkg upgrade-all'.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
The configuration file must be in the liveslak main directory ('/liveslak'
by default) and be called 'DISTRO_os.cfg' where DISTRO is the lower-case
name of the distro ('slackware' by default).
The following variables can be defined in the configuration file,
one per line, in the format VAR=value:
BLACKLIST, INIT, KEYMAP, LIVE_HOSTNAME, LOAD, LOCALE, LUKSVOL, NOLOAD,
RUNLEVEL, TWEAKS, TZ, XKB.
Since the configuration file is sourced by the live init script, it is OK
to have comments in it that are prefixed with '#'.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
The order in which the squashfs modules were created should be respected
when they are re-assembled to an overlay filesystem.
The assembly is done in lexicographical order - therefore xbase.lst
needed to be renamed to x_base.lst.
Failing to adhere to the original assembly order causes weirdness.
In my case it meant that the complete directory /usr/share/applications
as contained in xapbase.lst was invisible in the resulting overlay, causing
near-empty menus in XFCE.
Also added several packages to x_base.lst that are now part of Slackware.
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Group membership of 'input' is needed to access input devices that are
exposed through libinput (touch devices are only exposed through libinput)
and seems to be a requirement for Wayland.
|
| | |
|
| |
|
|
| |
ghostscript is again removed to make room for new additions in -current.
|
| |
|
|
|
| |
Falkon is the renamed qupzilla browser. Eventually, falkon will find
its way into Applications. Consider this a preview version.
|
| |
|
|
|
|
|
| |
- Explanation of the partition layout of a USB stick
when created with iso2usb.sh
- Updated URLS
- Many small fixes, mostly typos
|
| | |
|
| | |
|
| |
|
|
| |
There are no KDE4 packages left in the latest Plasma 5 releases.
|
| |
|
|
|
| |
frameworks: purpose
applications-extra: calligraplan
|
| | |
|
| |
|
|
|
|
|
|
| |
- acpilight is able to control the backlight brightness of a laptop screen,
this will be used by the Lumina Desktop if present.
- poppler-qt5 is a new dependency for Lumina Desktop in case you do not have
my Plasma5 'ktown' packages installed, specifically its poppler package
which already contains the required Qt5 libraries.
|
| |
|
|
|
|
|
|
| |
In particular, it seems that including the 'ahci' grub module caused failure
on boot, because the actual Live USB stick would not be detected and the
local harddisk would be detected as (ahci0).
The script has also been enhanced so that the 'grub-mkimage' commandline will
not get fed any non-existing module, which would otherwise abort the command.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
Eric Hameleers