sfs_load-2.4 on-the-fly

[jpeps]

I think you have in haste simply duplicated the last part of the first example?

not haste I ran the root shell as "bash" instead of "sh" to confirm that both act similarly.

Any idea why the behaviour is as you found it? The mystery has merely shifted focus.

Yes; it needs .wh\* to find it when not in root shell. For some reason, in root shell '*' doesn't need to be escaped.

Code: Select all

sudo find /initrd/pup_rw  -mindepth 2 -type f -name .wh\* 
/initrd/pup_rw/var/local/icons/.wh..wh..opq
/initrd/pup_rw/dev/usb/.wh.lp0
/initrd/pup_rw/etc/.wh.windowmanager.openbox
/initrd/pup_rw/etc/ssl/misc/.wh..wh..opq
/initrd/pup_rw/usr/lib/openoffice.org3/share/uno_packages/cache/uno_packages/.wh.SH1dMc_
/initrd/pup_rw/usr/lib/openoffice.org3/share/uno_packages/cache/uno_packages/.wh.SH1dMc
/initrd/pup_rw/usr/lib/engines/.wh..wh..opq

The dot is not a wildcard to the shell, so escaping it with a backslash achieves nothing.

[Q5sys]

[edit: Actually, I think this has more to do with running a user shell on top of the root shell. If I switch to sh shell with user "spot", it's the same as it was with bash.

And what exactly is the purpose of opening a shell to open a shell as you were doing? Am I missing something?

As far as the puppies ive used in the past including LHP currently... sh is a symlink to bash.

[Shep]

Yes; it needs .wh\* to find it when not in root shell. For some reason, in root shell '*' doesn't need to be escaped.

Hmm. Surely there isn't an environment setting such as expandwildcards=false
Though different for root and spot?

So we are no nearer understanding the apparent anomaly?

What about:

Code: Select all

 USER=root find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 
 USER=spot find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 

The dot is not a wildcard to the shell, so escaping it with a backslash achieves nothing.

A dot is not an asterisk.

[jpeps]

What about:

Code: Select all

 USER=root find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 
The dot is not a wildcard to the shell, so escaping it with a backslash achieves nothing.
 USER=spot find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 

no cigar

"The dot is not a wildcard to the shell, so escaping it with a backslash achieves nothing."

A dot is not an asterisk.

right...should have escaped the wildcard (although you can't use both quotes and '\*'). That said, dots are special characters in search patterns, so it's not a bad idea to get used to escaping them as well. Always better to write safe code to begin with that will cover all situations.

[Shep]

no cigar

Realised as soon as I'd posted that it wouldn't. The su command exists for a reason.
I'm trying to replicate the essence of your finding on Wary. Should be possible, you reckon?

So I get to a shell (shouldn't matter whether sh or bash, you say) and ....

Code: Select all

sh-3.00# echo $USER
root
sh-3.00# find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 
/initrd/pup_rw/lib/modules/all-firmware/.wh.cdcacm.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.hso.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.ipwireless.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.nozomi.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.usbserial.tar.gz
/initrd/pup_rw/dev/.wh.sr1
sh-3.00# su spot
sh-3.00$ echo $USER
spot
sh-3.00$ find /initrd/pup_rw  -mindepth 2 -type f -name .wh* 2>/dev/null
/initrd/pup_rw/lib/modules/all-firmware/.wh.cdcacm.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.hso.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.ipwireless.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.nozomi.tar.gz
/initrd/pup_rw/lib/modules/all-firmware/.wh.usbserial.tar.gz
/initrd/pup_rw/dev/.wh.sr1
sh-3.00$ 

The fact that I needed to discard lots of diagnostic messages while your spot didn't elicit any permission-denied messages gives pause for thought. Why was your spot able to descend into directories that presumably have root-only permissions? That's probably a more important question.

Any thoughts on that?

[Shep]

And what exactly is the purpose of opening a shell to open a shell as you were doing?

How would you have done it?

[jpeps]

The fact that I needed to discard lots of diagnostic messages while your spot didn't elicit any permission-denied messages gives pause for thought. Why was your spot able to descend into directories that presumably have root-only permissions? That's probably a more important question.

Any thoughts on that?

requires sudoers for root privileges:

http://murga-linux.com/puppy/viewtopic.php?t=60258

[Q5sys]

And what exactly is the purpose of opening a shell to open a shell as you were doing?

How would you have done it?

Um... I'm not understanding the reason of opening a shell to open a shell. Hence my question of what im missing. What is the reason you cant open one shell and issue the commands there? Whats the purpose of opening a nested shell inside the first one? Thats what Im not getting.

Nested shells make me think of the yo dawg meme...

[Shep]

Um... I'm not understanding the reason of opening a shell to open a shell.

It's more a case of first opening a console, then running the desired shell. The fact that the console defaults to opening with bash means that you need to get sh running if you want to test sh.

It could be done differently, but who cares? Shells are just programs, so for the case in question, it shouldn't be a problem. Finish with exit to return to bash.

Could use exec sh if you don't need to return to bash.

[Q5sys]

Um... I'm not understanding the reason of opening a shell to open a shell.

It's more a case of first opening a console, then running the desired shell. The fact that the console defaults to opening with bash means that you need to get sh running if you want to test sh.

It could be done differently, but who cares? Shells are just programs, so for the case in question, it shouldn't be a problem. Finish with exit to return to bash.

Could use exec sh if you don't need to return to bash.

Ok I ask, because Im hoping to further educate myself about this. So dont think i'm being a smart ass. lol.
What benefits are there to opening a shell within a shell versus using the exec command? In the past when I needed to open a shell and run something (like htop for a simple example) I've used exec. Is there any benefit to nesting shells?

ps... sorry to hijack your thread.

[jim3630]

shinobar, thanks for the pet it worked in my fatdog 64 511 without a hitch today. what a work saver! thanks again jim

[Shep]

What benefits are there to opening a shell within a shell versus using the exec command?

Well, there's less typing for a start.

No great drawback as far as I can see, provided you are not running low on memory.

It means that you can test a bit of code (e.g., by running it in various shells, ksh, zsh, dash, ash, bash, ...) then return to what you were doing. It's a straightforward way to inherit data, if that's desired. Inheritence goes only one way:

Code: Select all

# echo $0
bash
# echo $word

# export word
# word='Some important string or data'
# echo $word
Some important string or data
# sh
# echo $word
Some important string or data
# word='changed'
# echo $word
changed
# exit
exit
# echo $word
Some important string or data
# echo $0
bash
#

HTH

[mavrothal]

I see the big warning in v0.9 that in pupmode 6 is experimental, but it simply fails. "Failed to mount /initrd/pup_rw/name.sfs on /initrd/pup_roX" is all I get
v0.8 with the little addition was working OK...

[shinobar]

I see the big warning in v0.9 that in pupmode 6 is experimental, but it simply fails. "Failed to mount /initrd/pup_rw/name.sfs on /initrd/pup_roX" is all I get

Thanks mavrothal for the report, but i cannot reproduce your problem.
I tested sfs_load-0.9 with wary-503q on a virtual PC and successfully loaded sfs under PUPMODE=6...

[mhanifpriatama]

Just report. I try in my lucid puppy 520, fresh frugal install. And run.
I would be very happy if can click once, then run the program directly.
Like at http://portablelinuxapps.org

[amigo]

When you open a terminal it will be running whatever shell your default is set to. If you then start another shell, it will remain running until you type 'exit' whiuch returns you to the original shell.

There can be a great difference between /bin/sh and /bin/bash -even if /bin/sh is a link to /bin/bash. Running bash called as /bin/sh yupports *some* bash extennsions which are not supported by 'real' sh. if /bin/sh is really ash, or dash, or busybox ash/lash/hash/mash, then expect even more differences. None of the above are really 'sh' as it was in the old days -ash and dash support more features and any of the busybox versions will most likely support less. busybox ash is not the same as ash.

The point is, for consistency, you need to use the shebang for the exact shell which supports the code you are writing. If you use '#!/bin/sh' as the shebang, don't assume that /bin/sh will be bash and act like bash-as-sh. if you really want to use /bin/sh, then you'll have to check all of your syntay to make sure it works with whatever /bin/sh really is. Once again, if you call bash as 'bash' it will (obviously) support all the bash features which were compiled into it. But, if you run bash as 'sh', it will only support a small sub-set of the bash extensions.

[shinobar]

Just report. I try in my lucid puppy 520, fresh frugal install. And run.
I would be very happy if can click once, then run the program directly.
Like at http://portablelinuxapps.org

If the sfs is properly prepaired, you can right-click the sfs and select 'sfs_load'. After loading succeed, the launcher offers to run the program.
Try the gimp sfs for an example.
http://www.murga-linux.com/puppy/viewtopic.php?t=61584

[01micko]

Hi shino (good to see you are ok )

What about an adjustment to the run action for an sfs.. depending upon sfs_load's existence?

/root/Choices/MIME-types/application_x-squashfs-image

Code: Select all

#! /bin/sh
. /etc/rc.d/BOOTCONFIG
FILE="`basename $1`"
IN_USE=`echo $LASTUNIONRECORD|grep "$FILE"`
[ "$IN_USE" != "" ]&& xmessage -bg orange -c -timeout 5 "$FILE 
is already in use by Puppy" && exit
MOUNTED=`mount|grep "$FILE"`
[ "$MOUNTED" != "" ]&& exec filemnt "$1"
if [ -x /usr/sbin/sfs_load ];then 
 xmessage -bg green -c -buttons "View only:10,Load to Use:11,Exit:12" "Would you like to mount 
 $FILE
 for viewing read only or 
 mount for use with sfs_load?"
 case "$?" in
 10) exec filemnt "$1" ;;
 11) sfs_load "$1" ;;
 12) exit ;;
 esac
 else
 exec filemnt "$1"
fi

Of course you could use Xdialog or gtkdialog if you wish.

Cheers

EDIT: On further thought, if this were to be an approach, then really /usr/sbin/filemnt should get the rewrite.

[lithpr]

This is fantastic. works flawless under 2.5.4, as far as i can tell. Thank you!

Lithpr

[mhanifpriatama]

Shinobar :

If the sfs is properly prepaired,

Ohh, what is that? I don't know that.
Sorry.
All of sfs files must properly prepaired?

I'm fall-in-love to fresh frugal install and remastering.
Maybe we can create a standard of sfs file? Or, create a new file type, that similar to portable applications?
I use portable things (some worked, some didn't), but there're a lot of pet's packages that can convert to sfs easily.
Do not mind me, if being too demanding.