Persistent Unloaded SFS

[JASpup]

This is not a request for advice but a process bug:

If I SFS-Load-on-the-fly to spot, unloading the app appears to work but the app is still present. There's a warning about restarting-X, but it persists.

App does not appear loaded in the gui after unloading, but can still be run. I wrote about this for a different system/app here: viewtopic.php?f=2&t=1637&p=12285&hilit=Opera#p12285 (recurring bug)

The cincher is the SFS will go to your pupsave like a normal installation.

64Xenial XFCE
SFS-Load v3.0

[s243a]

That sounds like a bug. Please provide some details such as:
1. what version of puppy are you using
2. how did you unload the sfs
3. how did you load the sfs?

[bigpup]

SFS-Load-on-the-fly to spot

,
Explain this????

Normally all sfs packages are placed in /mnt/home
They are loaded from there.
/mnt/home is completely outside of the save.

If this is in Xenialpup
Need to do the Xenialpup updates using Quickpet->Info->Xenialpup updates.

[taersh]

This is NOT a bug.

This may happen if another process from the program of the .sfs is still running after the program was quit.
E.g.: the .sfs from KdenLive video editor will NOT unload after the program was quit and trying to unload the .sfs.
This is caused by processes klogd klauncher kdeinit5 still running. If these processes are killed, the .sfs unloads fine.

The .sfs disappears from the list though, and even restarting X will not unload the .sfs in some cases. Usually it's unloaded after a reboot.

So, it would be good to know the name of the .sfs or better the name of the program of the .sfs.
Then there could be a workaround/solution.

[JASpup]

I could recreate the problem, but it was such a mess it begs a testing scenario.

I boot the XFCE Xenial on SourceForge by BattleShooter.

The SFS is Google_Chrome-63.0.3239.84-amd64-tahr.sfs.

It runs as spot via a script.

I've loaded by clicking in file manager (ROX function is the same), yet I think what happened this time is that didn't work because SFS-Load didn't like the special characters, so I loaded from within the gui.

Once a SFS appears in the gui, that's usually how I unload it.

That sounds like a bug. Please provide some details such as:
1. what version of puppy are you using
2. how did you unload the sfs
3. how did you load the sfs?

[JASpup]

SFS-Load-on-the-fly to spot

,
Explain this????

Normally all sfs packages are placed in /mnt/home
They are loaded from there.
/mnt/home is completely outside of the save.

If this is in Xenialpup
Need to do the Xenialpup updates using Quickpet->Info->Xenialpup updates.

I appreciate your stressing updates, though I like confidence that it's the cause of an error (e.g., neglected updates didn't cause my partition click problem).

Commonly these skami browsers are being loaded to spot. If they're not unloaded, that's how they wind up in a pupsave.

If you issue a command to unload a SFS, it should be unloaded no matter where it resides, and if it remains loaded, it should display as such.

Presuming you use the app, there's an option of copying the SFS to /mnt/home or not. If that's a technical issue a warning would help. I've noticed, it fills disk space, but I'm more interested in the load destination.

[JASpup]

This would be technical insight for developers, but when a user is using an app as intended and it exhibits these problems, it's not user error.

The app would need to be more error-proof, if it's worth anyone's time.

This is NOT a bug.

This may happen if another process from the program of the .sfs is still running after the program was quit.
E.g.: the .sfs from KdenLive video editor will NOT unload after the program was quit and trying to unload the .sfs.
This is caused by processes klogd klauncher kdeinit5 still running. If these processes are killed, the .sfs unloads fine.

The .sfs disappears from the list though, and even restarting X will not unload the .sfs in some cases. Usually it's unloaded after a reboot.

So, it would be good to know the name of the .sfs or better the name of the program of the .sfs.
Then there could be a workaround/solution.

[s243a]

I could recreate the problem, but it was such a mess it begs a testing scenario.

I boot the XFCE Xenial on SourceForge by BattleShooter.

The SFS is Google_Chrome-63.0.3239.84-amd64-tahr.sfs.

It runs as spot via a script.

I've loaded by clicking in file manager (ROX function is the same), yet I think what happened this time is that didn't work because SFS-Load didn't like the special characters, so I loaded from within the gui.

Once a SFS appears in the gui, that's usually how I unload it.

That sounds like a bug. Please provide some details such as:
1. what version of puppy are you using
2. how did you unload the sfs
3. how did you load the sfs?

From taersh post above, I think that the unload_sfs command could have an option for interactive mode. In this mode, when the user unloads an sfs it will check if any process are running which depend on the sfs and if there are any such processes then the unload_sfs function will kill them for you (if you choose), prior to unloading the sfs.

[JASpup]

From taersh post above, I think that the unload_sfs command could have an option for interactive mode. In this mode, when the user unloads an sfs it will check if any process are running which depend on the sfs and if there are any such processes then the unload_sfs function will kill them for you (if you choose), prior to unloading the sfs.

That would make sense, but the gui is not that sophisticated.

[s243a]

From taersh post above, I think that the unload_sfs command could have an option for interactive mode. In this mode, when the user unloads an sfs it will check if any process are running which depend on the sfs and if there are any such processes then the unload_sfs function will kill them for you (if you choose), prior to unloading the sfs.

That would make sense, but the gui is not that sophisticated.

We're all volunteers so someone would have to make these changes and do a pull/merge request. You could open an issue about this on github. I can't promise good reception, but who knows. Maybe the devs will like the idea.

[taersh]

From taersh post above, I think that the unload_sfs command could have an option for interactive mode. In this mode, when the user unloads an sfs it will check if any process are running which depend on the sfs and if there are any such processes then the unload_sfs function will kill them for you (if you choose), prior to unloading the sfs.

That's exactly what I'm doing within my RunScripts. Though, it's not done automatically. If I'd found a .sfs module that won't unload, I'd searched for the programs from that .sfs module via taskmanager. So, I found klogd klauncher kdeinit5 still running after exiting KdenLive. In older versions of KdenLive it was kdeinit4 instead of kdeinit5.

In general I'm loading .sfs modules by a RunScript. All my programs from sfs. modules have a menu entry in my system. When I'm clicking this menu entry, e.g. GIMP, it won't run the GIMP, since GIMP is NOT installed.

Instead the menu entry (the Exec= field inside the .desktop file) is calling the GIMP-RunScript. This RunScript loads the .sfs module of GIMP, after successfully loaded it executes GIMP and a few seconds later it creates an Unload-Script to automatically unload the GIMP .sfs module after exiting GIMP.

Inside of the RunScript there's some fields/parameters to set up. One of these entries is for killing processes which can have multiple entries for programs to be killed before trying to unload the .sfs module.

If a program like e.g. JWildFire needs a dependency to load -in this case: JAVA- there's also a field to have multiple entries for dependent .sfs modules to load before executing the program from the loaded e.g. JWildFire .sfs module. Also I can define .sfs modules to be unloaded first in case they would get in conflict with the .sfs module to be loaded. For some years I used this to unload either Openshot before loading KdenLive and revert. Meanwhile I can load and use both of them without getting in conflict to each other.

Attached is my RunScript for my KdenLive .sfs for examination. It's a fake .gz file.
It's well commented, I think, though one needs to have a little bash experience to make it working for own needs.

It's exactly that stuff one can find and use by the SFS PLUS 2 application.

run-kdenlive.gz

(6.12 KiB) Downloaded 34 times

[wanthinker]

This is not a request for advice but a process bug:

If I SFS-Load-on-the-fly to spot, unloading the app appears to work but the app is still present. There's a warning about restarting-X, but it persists.

App does not appear loaded in the gui after unloading, but can still be run. I wrote about this for a different system/app here: https://forum.puppylinux.com/viewtopic. ... era#p12285 (recurring bug)

The cincher is the SFS will go to your pupsave like a normal installation.

64Xenial XFCE
SFS-Load v3.0

I had this happen once before. Usually the thing i do to prevent this somewhat is:

- copy the sfs to a directory, usually /tmp/ (well aware you technically can't install sfs in /tmp/ but bear with me)
- cd in /tmp/ (very important to bypass the "cannot install in /tmp/)
- load it like so:

Code: Select all

sfs_loadr -s -q -c sfsfile.sfs

- remove the sfs right then and there after it's loaded
- done

On shutdown, even if you didn't unload it, it shouldn't be part of your save file (at least this shouldn't if it's apps that reside in /usr/ etc).
The app should work pretty well too, tested this with devx, qemu and many other sfs.

[JASpup]

I'll check this out next time I'm on the issue.

The only thing occurring recently is Opera SFS not unloading despite being closed in Tahr 32.

Not sure the list of technical hiccups that prevent unloading SFS, but fortunately it does not remove the entry from SFS-load.

This is not a request for advice but a process bug:

If I SFS-Load-on-the-fly to spot, unloading the app appears to work but the app is still present. There's a warning about restarting-X, but it persists.

App does not appear loaded in the gui after unloading, but can still be run. I wrote about this for a different system/app here: viewtopic.php?f=2&t=1637&p=12285&hilit=Opera#p12285 (recurring bug)

The cincher is the SFS will go to your pupsave like a normal installation.

64Xenial XFCE
SFS-Load v3.0

I had this happen once before. Usually the thing i do to prevent this somewhat is:

- copy the sfs to a directory, usually /tmp/ (well aware you technically can't install sfs in /tmp/ but bear with me)
- cd in /tmp/ (very important to bypass the "cannot install in /tmp/)
- load it like so:

Code: Select all

sfs_loadr -s -q -c sfsfile.sfs

- remove the sfs right then and there after it's loaded
- done

On shutdown, even if you didn't unload it, it shouldn't be part of your save file (at least this shouldn't if it's apps that reside in /usr/ etc).
The app should work pretty well too, tested this with devx, qemu and many other sfs.

[bigpup]

This is not a request for advice but a process bug:

If I SFS-Load-on-the-fly to spot, unloading the app appears to work but the app is still present. There's a warning about restarting-X, but it persists.

App does not appear loaded in the gui after unloading, but can still be run. I wrote about this for a different system/app here: viewtopic.php?f=2&t=1637&p=12285&hilit=Opera#p12285 (recurring bug)

The cincher is the SFS will go to your pupsave like a normal installation.

64Xenial XFCE
SFS-Load v3.0

If I SFS-Load-on-the-fly to spot

If this is saying the sfs package is stored in spot directory.
That is in the save.
The Puppy expected location is /mnt/home for SFS packages to be stored.
That is outside of the save.

Is this the type of warning message you get?
.

Screenshot(2).png (14.96 KiB) Viewed 591 times

.
If yes.
Whatever the program is in the SFS package.
You may shutdown the program normally, but it is not totally shutdown.
The program probably has some part of it that is still loaded in RAM and still running.

What is the specific SFS package and a link to download and try it?

Just because it is an SFS package does not make it correctly put together, to load, run, and unload in Puppy Linux.

[JASpup]

Yes, I am not experiencing this error daily (nor at the moment), but I understand how technical details can be missed in building the SFS that can affect usability.

If I SFS-Load-on-the-fly to spot

If this is saying the sfs package is stored in spot directory.
That is in the save.
The Puppy expected location is /mnt/home for SFS packages to be stored.
That is outside of the save.

Is this the type of warning message you get?
.
Screenshot(2).png
.
If yes.
Whatever the program is in the SFS package.
You may shutdown the program normally, but it is not totally shutdown.
The program probably has some part of it that is still loaded in RAM and still running.

What is the specific SFS package and a link to download and try it?

Just because it is an SFS package does not make it correctly put together, to load, run, and unload in Puppy Linux.