- Slackware
- Ubuntu
- Debian
Puppy can be built directly on GitHub, saving the hassle of preparing a suitable build environment on a fast machine with adequate storage.
-
Fork woof-CE.
-
Clone your fork.
-
Modify woof-CE, commit your changes and push them.
-
Download your Puppy from the newly published release.
-
Test your Puppy and open a pull request to woof-CE, if you want your changes to be officially incorporated into woof-CE.
To modify woof-CE and push the changes to GitHub without having to set up a local development environment:
Not all of woof-CE can run without root access, but most of woof-CE works on Gitpod and it's possible to boot Puppy inside QEMU and control it over VNC, all through the browser.
To work on woof-CE in a fork, through Gitpod:
-
Fork woof-CE.
-
Browse to
https://gitpod.io/#https://github.com/your-github-username/woof-CE
.
Woof-CE has five directories:
- woof-arch : architecture-dependent (x86_64, x86, ARM) files, mostly binary executables.
- woof-code : the core of Woof. Mostly scripts.
- woof-distro : distro-configuration (Debian, Slackware, etc.) files.
- kernel-kit : scripts to download, patch, configure and build the kernel.
- initrd-progs: scripts and files to generate the initial ramdisk
To create a working directory, named woof-out_*
, you first have to run the merge2out
script. This merges the 5 directories into a directory named woof-out_*
. You then cd
into woof-out_*
and run the build scripts.
The great thing about this merge operation is that you can choose exactly what you want to go into woof-out. You can choose the host system that you are building on (usually x86_64), the target (exs: x86_64 x86, ARM), the compatible-distro (ex: slackware), and the compat-distro version (ex: 15.0). So, you create woof-out without any confusing inappropriate content.
So, to get going with woof-CE, open a terminal and do this:
./merge2out
cd ../woof-out_*
- Suitable build environment
- Linux partition (ext2/3/4)
- At least 6-10GBs of space
- Host operating system
- A recent Woof-CE puppy with the devx (compilers, headers and other development tools) installed. Otherwise use run_woof.
- Choose a compatible-distro.
This is the distro whose packages you are going to 'borrow' to build your Puppy. Open file DISTRO_SPECS in a text editor and change this line:
DISTRO_BINARY_COMPAT="ubuntu"
to what you want: slackware
, devuan
, ubuntu
, debian
or puppy
.
Open a terminal in the woof-out_*
directory.
-
Download package database files
./0setup
OPTIONAL: Tweak common PET package selection. You can edit the variable PKGS_SPECS_TABLE in file DISTRO_PKGS_SPECS-*
to choose the packages that you want in your build.
-
Download packages
./1download
About 500MB drive space is required, but this may vary enormously depending on the package selection.
-
Build the cut-down generic Puppy-packages
./2createpackages
-
Build Puppy live-CD
./3builddistro
This gets built in a directory named sandbox3
and as well as the live-CD ISO file you will also find the individual built files and the devx
file.
Woof-CE (woof-Community Edition) is a fork of Barry Kauler's woof2 fossil repository of Nov 11, 2013 commit f6332edbc4a75c262a8fec6e7d39229b0acf32cd.
any directory in the template, the files in the target pkg will be cut down to the same selection (even if empty dir). Exception, file named PLUSEXTRAFILES
then target will have all files from deb.
- 0-size file, means get file of same name from deb (even if in different dir) to target.
- Non-zero file, means copy this file from template to target.
- Template files with
-FULL
suffix, rename target file also (exs: in coreutils, util-linux).
Any dir in template with PKGVERSION
in name, substitute actual pkg version number in target dir. Except for /dev, /var, all dirs in target are deleted to only those in template, except if file PLUSEXTRADIRS
is found in template.
As a last resort, if target pkg is wrong, a file FIXUPHACK
is a script that can be at top dir in template. It executes in target, with current-dir set to where FIXUPHACK
is located. (ex: perl_tiny). Ran into problem slackware post-install scripts messing things up. See near bottom of '2createpackages' how damage is limited. Also DISABLE_POST_INSTALL_SCRIPT=yes
in FIXUPHACK
to disable entirely.
If a dir in template has files in it then target is cut down (unless PLUSEXTRAFILES
present), however there are some exceptions (such as .so regular files).
Notice that there are Packages-puppy-noarch-official
, also Packages-puppy-common-official
The single-digit -2-
, -3-
, -4-
, -5-
files reside on ibiblio.org also. These files list the complete contents of each repository.
The main Puppy files are:
vmlinuz, initrd.gz, puppy.sfs, zdrv.sfs, fdrv.sfs, adrv.sfs, ydrv.sfs
Versioning is put into the last two, for example:
vmlinuz, initrd.gz, puppy_slacko_7.0.0, zdrv_slacko_7.0.0.sfs fdrv_slacko_7.0.0.sfs, adrv_slacko_7.0.0.sfs, ydrv_slacko_7.0.0.sfs
...those last two names are intended to be unique for that build of Puppy, so they can be found at bootup.
The filenames are stored in the built Puppy, in /etc/DISTRO_SPECS. For example:
DISTRO_PUPPYSFS='puppy_slacko_7.0.0.sfs'
DISTRO_ZDRVSFS='zdrv_slacko_7.0.0.sfs'
DISTRO_FDRVSFS='fdrv_slacko_7.0.0.sfs'
DISTRO_ADRVSFS='adrv_slacko_7.0.0.sfs'
DISTRO_YDRVSFS='ydrv_slacko_7.0.0.sfs'
So, any script that wants to know what the names are can read these variables.
Woof 3builddistro also copies DISTRO_SPECS into the initrd.gz, so that the init
script can see what files to search for.
However, in a running Puppy, you can find out the filenames in the way that scripts have done before, by reading PUPSFS
and ZDRV
variables in /etc/rc.d/PUPSTATE.
In fact, to clarify the difference between these two sets of variables, I have put this comment into /etc/DISTRO_SPECS:
#Note, the .sfs files below are what the `init` script in initrd.gz searches for,
#for the partition, path and actual files loaded, see `PUPSFS` and `ZDRV` in /etc/rc.d/PUPSTATE
Woof-CE, a fork of woof2 can build the same as woof2 however a new feature has been added as of today. It now has the ability to build a distro with out modules in the initrd.gz, a feature which had been pioneered by Fatdog developers kirk and jamesbond. This has a number of advantages over the legacy kernel builds.
- No messy copying kernel modules from the initial ram disk to the root system.
- Ease of changing kernels.
I have named this the 'huge' type kernel, for want of a better term. The rationale for this is that Slackware developers name their default kernel huge-$some_suffix
. The reason is that the vmlinuz kernel image contains all the necessary filesystem and hardware drivers to get the system to boot and hand over to the real running system. Once that occurs, kernel modules are loaded to bring up the rest of the hardware and extra filesystems if necessary.
"kernel-kit", part of woof-CE, has the ability to produce one of these 'huge' style kernel packages. Please read the relevant README and the comments in "build.conf" inside the kernel-kit directory.
If you have built a "huge" style kernel with kernel-kit then place the package in the "huge_kernel" directory at the root of your woof installation. If not, one will be downloaded for you after you invoke 3builddistro from the CLI. You do get a choice of which version you want. Be sure you choose the correct architecure. All 32 bit builds are suffixed with either i486, i686 or x86. All 64 bit builds are suffixed x86_64. At the end you will end up with an ISO image, devx and checksums as usual.
Regards, Barry Kauler puppylinux.com