Is there interest in an Arch32 Pup???

[peebee]

Having got Arch32Pup this far I would like somebody else to take it the next steps to Beta/RC quality.....

If somebody does volunteer, I will prepare a tarball to add all the necessary bits and pieces to a Woof-CE build.

Step forward and form a queue??!!

p.s. an Arch64Pup is also a possibility having got this far.....

[rockedge]

I have a lot on the plate ...but I could take a look at the tarball....make a couple of versions and see if I can get some work done on further development.

It does seem to have potential

[s243a]

I have a lot on the plate ...but I could take a look at the tarball....make a couple of versions and see if I can get some work done on further development.

It does seem to have potential

If you want to do the building I could maintain a github or gitlab repo for arch32pup which you would have commit rights. Of course we would try to push as much as possible to woof-CE official.

I suppose, I also could try building via woof-CE but I've never done so and I think that it takes a long time. I would like to focus more on the code than the building

P.S. Regarding the kernels, I would like there to be one with a 64bit kernel and one with a 32bit kernel.

I am thankful, for all the work the Peebee has done here and would miss the help if Peebee steps away from this project but I understand that Peebee has a lot of things on Peebee's plate and might want the time to focus on other projects.

P.S. rockedge, if we go this way, then please let me know any changes you make outside of github/gitlab so we can consider how to include them in github/gitlab.

[rockedge]

I have a fork of woof-CE here : https://github.com/techrockedge

I will see if I can finish up one or two projects that are nearing completion today and then try a woof-CE build of ArchPup to get into the swing of things.....

I could use a fork of ArchPup as well....

[darry19662018]

13-March-2020

Arch32Pup-20.03+0.iso (md5=d33a721469412d1c9292072735346824)

Kernel 4.14.173 32-bits pae
Woof-CE testing;fbe454968;2020-03-10 15:25:12
Arch32Linux updates

Download links:
http://www.smokey01.com/peebee/downloads/a32pup/

inc. devx, kernel-headers and kernel-sources in othersfs

Still very Alpha quality....see next post

Thanks Peebee

[s243a]

I have a fork of woof-CE here : https://github.com/techrockedge

I will see if I can finish up one or two projects that are nearing completion today and then try a woof-CE build of ArchPup to get into the swing of things.....

I could use a fork of ArchPup as well....

Do you have Peebee's tarball?

[s243a]

Having got Arch32Pup this far I would like somebody else to take it the next steps to Beta/RC quality.....

If somebody does volunteer, I will prepare a tarball to add all the necessary bits and pieces to a Woof-CE build.

Step forward and form a queue??!!

p.s. an Arch64Pup is also a possibility having got this far.....

I'll give the Arch64Pup a try since rockedge want's to try building arch32pup. So may I please have the tarball, or better yet, why not just upload it to something like github or gitlab? If you you don't want to maintain it you can give one or more people commit rights.
...
thinking a bit more...you said that you had to compile stuff? What might I have to compile for arch64pup and what tools might I need to do the compiling?

[s243a]

I have a fork of woof-CE here : https://github.com/techrockedge

I will see if I can finish up one or two projects that are nearing completion today and then try a woof-CE build of ArchPup to get into the swing of things.....

I could use a fork of ArchPup as well....

If you upload peebee's tarball, I'll send you pull/merge requests for any improvements I make. .

[s243a]

I discovered that if arch32pup is the host and I ssh into the host than chroot into a another arch32pup, I can X11 forward as follows:

Code: Select all

mount --bind /tmp/.X11-unix /mnt/sb/fakeroot/tmp/.X11-unix
export DISPLAY=:10

Then the chroot version of arch32pup can talk to the xserver created the ssh server, when X11 forwarding is used. This is necessary on arch32pup because on arch the .Xauthority file created by the ssh server give only premission for sockets to access the ssh server.

The following command shows that the X11 server is bound to and ip address:

Code: Select all

netstat -plnt
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       
tcp        0      0 0.0.0.0:6010            0.0.0.0:*               LISTEN nt

but if I do xauth -list

Code: Select all

xauth list
puppypc1365/unix:10  MIT-MAGIC-COOKIE-1 SOME-COOKIE-HASH

I see that users are only granted permission to the X11 server via sockets (TCP access not allowed). If josejp2424 dpup buster as the host is used then the xauth will allow tcp access.

The arch approach is more secure but it means that I have to bind tmp/.X11-unix into the chroot. Not sure what settings maybe differ between dpup buster and arch32 pup for this difference.

On another note, I notice that when dpup buster is the host the chroot inherits the host name but not when arch32pup is the host. On arch32 pup if you want to inherit the host name then you have to copy /etc/hostname into the chroot.

[s243a]

This post and last testing the latest 20.03:

After installing firefox via the ppm, I still get this error:

Code: Select all

sandbox# firefox
XPCOMGlueLoad error for file /usr/lib/firefox/libxul.so:
libmozsandbox.so: cannot open shared object file: No such file or directory

to fix the error add the following line to the end of "/etc/ld.so.conf"

Code: Select all

include /etc/ld.so.conf.d/*.conf

then run the command:

Code: Select all

ldconfig

as previously metioned I have a pull request related to this on woof-CE:
in ld.so.conf add: include /etc/ld.so.conf.d/*.conf #1785

[peebee]

I haven't made the tarball yet.........next couple of days........will upload it when complete

Yes - the ffmpeg app needed recompile - devx + kernel headers and sources needed

You should always use the Woof-CE Testing Branch to build to be sure of getting latest

A build takes about 5 mins on my 8-core desktop with ssd "hard" drive. Half of that is the mksquashfs stage,

You can do incremental builds ( ./2createpackages -downloaded ) which makes things much faster

[peebee]

Tarball and instructions:

http://www.smokey01.com/peebee/downloads/test/

Sort of tested but there may be something that I've forgotten....

[rockedge]

Thanks! I have downloaded and setup a github repo

https://github.com/techrockedge/archpup32

I can provide commits, push access for you s243a and peebee if needed.

once a 64 bit is under way I will fork a version as well

[s243a]

Thanks! I have downloaded and setup a github repo

https://github.com/techrockedge/archpup32

I can provide commits, push access for you s243a and peebee if needed.

once a 64 bit is under way I will fork a version as well

Yes, please give me commit rights. I may also set up my own repo but either way I'll send you commits.

[rockedge]

Hello s243a,

I have sent an invite for collaboration to you

[rockedge]

I cloned a woof-ce-testing
ran merge2out
first run I made it to ./1download since then this error appears after another 6 attempts to run

Code: Select all

root-#  ./0setup
Packages-puppy-slac 100%[===================>]  65.62K   312KB/s    in 0.2s    
Packages-puppy-noar 100%[===================>]  31.63K  --.-KB/s    in 0.1s    
Packages-puppy-comm 100%[===================>]  20.34K  --.-KB/s    in 0.06s   
core.db.tar.gz      100%[===================>] 142.53K   313KB/s    in 0.5s    
Processing ArchLinux32 core repo please wait...
extra.db.tar.gz     100%[===================>]   1.90M   197KB/s    in 11s     
Processing ArchLinux32 extra repo please wait...
community.db.tar.gz 100%[===================>]   4.75M  97.2KB/s    in 36s     
Processing ArchLinux32 community repo please wait...
cat: sandbox0/archtemp/desc: Not a directory
root-#

the script bails out after the error.

[s243a]

I cloned a woof-ce-testing
ran merge2out
first run I made it to ./1download since then this error appears after another 6 attempts to run

Code: Select all

root-#  ./0setup
Packages-puppy-slac 100%[===================>]  65.62K   312KB/s    in 0.2s    
Packages-puppy-noar 100%[===================>]  31.63K  --.-KB/s    in 0.1s    
Packages-puppy-comm 100%[===================>]  20.34K  --.-KB/s    in 0.06s   
core.db.tar.gz      100%[===================>] 142.53K   313KB/s    in 0.5s    
Processing ArchLinux32 core repo please wait...
extra.db.tar.gz     100%[===================>]   1.90M   197KB/s    in 11s     
Processing ArchLinux32 extra repo please wait...
community.db.tar.gz 100%[===================>]   4.75M  97.2KB/s    in 36s     
Processing ArchLinux32 community repo please wait...
cat: sandbox0/archtemp/desc: Not a directory
root-#

the script bails out after the error.

You could get rid of the error by replacing

Code: Select all

 for i in sandbox0/*; do
  cat "$i/desc"

0setup#L35

with:

Code: Select all

 for i in sandbox0/*; do
  [ -d "$i/desc" ] && cat "$i/desc"

See:
https://linux.die.net/man/1/test

**However, this probably is a be a bad idea . I'll explain in another post.

Note, I haven't tried building yet but it looks the same 0setup script that is used to update the repo db/doc files in arch32pup.

There arch repo metdata is extracted to sandbox0. Each folder represents a package and the desc file gives the metadata for the package. You can trace the script like this:

bash -x ./0setup

You can also try the script in a running version of arch32pup to see if it is updating the repos properly. I'll try running the script to see what happens and to see if I can troubleshoot the error.

[s243a]

I cloned a woof-ce-testing
ran merge2out
first run I made it to ./1download since then this error appears after another 6 attempts to run

Code: Select all

root-#  ./0setup
Packages-puppy-slac 100%[===================>]  65.62K   312KB/s    in 0.2s    
Packages-puppy-noar 100%[===================>]  31.63K  --.-KB/s    in 0.1s    
Packages-puppy-comm 100%[===================>]  20.34K  --.-KB/s    in 0.06s   
core.db.tar.gz      100%[===================>] 142.53K   313KB/s    in 0.5s    
Processing ArchLinux32 core repo please wait...
extra.db.tar.gz     100%[===================>]   1.90M   197KB/s    in 11s     
Processing ArchLinux32 extra repo please wait...
community.db.tar.gz 100%[===================>]   4.75M  97.2KB/s    in 36s     
Processing ArchLinux32 community repo please wait...
cat: sandbox0/archtemp/desc: Not a directory
root-#

the script bails out after the error.

I tried it and it worked for me. Maybe have a look at peebee's instructions again. My understanding is that the the tarball is meant to replace certain files in the output of woof-CE. Pebee said to rename the woof-CE output to woof-out_arch32. Then you put the tarball into this folder and extract it in order to replace certain files...while there was also some deletions before that.

Anyway, run the 0setup script again in this folder and see what happens.

If it doesn't work I'll try to help. Note that if we don't like the tarball approach, we could use a command like cpio instead to copy the folder contents from your repo into the woof-CE output...locally or course.

[rockedge]

odd thing is I had it working and it began to error out after download failures from the arch mirror url occurred. Since then I am seeing that error.

I have deleted the renamed woof-out_arch32 several times and re -ran merge2out to create it fresh. Then I would rename it and I've followed the instructions to a "T"

I am tracking down why this is occurring and why the downloads of certain packages from the pentium4 directory in the arch repo are failing to be found.

I have a list from the logs of the packages that are not found

I am attempting to build with a Bionic64-v8, I am switching over to a 32 bit Bionic

[s243a]

odd thing is I had it working and it began to error out after download failures from the arch mirror url occurred. Since then I am seeing that error.

I have deleted the renamed woof-out_arch32 several times and re -ran merge2out to create it fresh. Then I would rename it and I've followed the instructions to a "T"

I am tracking down why this is occurring and why the downloads of certain packages from the pentium4 directory in the arch repo are failing to be found.

I have a list from the logs of the packages that are not found

I am attempting to build with a Bionic64-v8, I am switching over to a 32 bit Bionic

The 0setup script is such that if it fails to process the archive, then it doesn't delete the archive. This is done so that you can inspect the contents of the archive to see what went wrong. The problem is that if there is a problem is with the download then you will keep using the old failed download rather than getting a fresh download.

I had this issue before. See post:
http://murga-linux.com/puppy/viewtopic. ... 04#1052004

also mentioned in the post is that the busybox tar command might be more robust to these kinds of errors. My main suggestion is to delete any archives that were downloaded prior to running the script (e.g. community.db.tar.gz ). However, to avoid this error in the fusture:

replacing:

Code: Select all

tar xf ../community.db.tar.gz 

with

Code: Select all

busybox tar xf ../community.db.tar.gz 

might help