Xenialpup CE 7.5 / 25 Nov 2017

[RetroTechGuy]

And, or course, the size of the save file. The whole save file needs to be written on shutdown

Not sure where you are getting this idea, but it is wrong.

After you first make the original save.
The only thing that gets written to the save, is anything that is not already in the save.
Stuff is just added to the save that is already there.

Hmmm... That's interesting. Then the slow shutdown write issue makes no sense at all.... I assumed that since the SFS had been changed, that the system didn't bother to try figuring out which compressed bits to save, and just rewrote the whole thing. What magic is it doing to keep track of which parts of the SFS need to be updated?

[bigpup]

How Puppy works:
http://bkhome.org/archive/puppylinux/de ... works.html
Look at pupmode 13 operation.

Then the slow shutdown write issue makes no sense at all.

The format of the USB flash drive could be an issue.
If it is fat 32 format and that file system is corrupted or probably fragmented.
Can happen with fat 32.

That would slow things down.

[RetroTechGuy]

How Puppy works:
http://bkhome.org/archive/puppylinux/de ... works.html
Look at pupmode 13 operation.

So, the compromise is that the contents of the top layer is copied down to the next layer -- that is officially supported by unionfs.

If it's really doing a copy, that perhaps still needs to push the entire contents (or rewrite the entire contents) of the save file.

I don't know enough about the guts of this. But my observation is that the write time is consistent with writing the entire save file, through whatever port speed limitations you suffer.

[6502coder]

The only thing that gets written to the save, is anything that is not already in the save.
Stuff is just added to the save that is already there.

I'm confused about this too. I understand that only new changes are added to the save file, but while the save file contains an entire file system internally, the save file ITSELF is just an ordinary file, as far as the OS is concerned. So doesn't the entire save file have to be written afresh to the storage medium if any part of it gets modified? It's not like the old days when we used to use a floppy disk "hex" editor to do direct read/writes to specific sectors on the disk. Can Linux actually update in-place only PART of a file on a storage medium?

[rcrsn51]

Can Linux actually update in-place only PART of a file on a storage medium?

Yes. This is how random-access databases work.

[mikeslr]

How Puppy works:
http://bkhome.org/archive/puppylinux/de ... works.html
Look at pupmode 13 operation.

Actually, that explains a lot once you factor in that at bootup Puppies create in RAM inodes ("pointers") to where the rest of the files --those not already copied into RAM-- are located.

Suppose you have application XYZ built into your Puppy. On bootup some of XYZ's files are copied into RAM and inodes created pointing to the rest of XYZ's files are located within Puppy_XXX_VersionNumber.sfs. When you install a new version of XYZ, initially that "install" only is written to /initrd/mnt/tmpfs/pup_rw. Barry K's posts shows /initrd/pup_rw but under current Puppies that is a symbolic link to the folder I mentioned. If you restart-x, the inodes will be re-cataloged and will now point to the XYZ files in /initrd/mnt/tmpfs/pup_rw. That, however, is still only in RAM. When a Save is executed the contents of /initrd/mnt/tmpfs/pup_rw are written to your SaveFile/Folder. On bootup, the files of your SaveFile/Folder have priority over the files in Puppy_XXX_VersionNumber.sfs so the inodes created will point to your SaveFile/Folder for the XYZ application.

[Of course, if you do not Save, RAM is cleared on shutdown/reboot so on reboot nothing previously in the /initrd/mnt/tmpfs/pup_rw file will be present and the inodes will still point to XYZ in the Puppy_XXX_VersionNumber.sfs].

If you look at the /initrd/mnt/tmpfs/pup_rw folder, you'll see that it contains folders corresponding to each of those at ~ (top-level Root; not /root). With two exceptions, each of those folders has, or can have created, within it subfolders corresponding to those folders with subfolder directly hanging from ~. The two exceptions are mnt --which on bootup only includes a symbolic link to /mnt/home but which may include others if you mount other media -- and tmp which is a symbolic link to /initrd/mnt/tmpfs/tmp. It is my understanding that Puppies are constructed so that files written to the latter will not be written to a SaveFile/Folder; and files written to the mnt will follow a chosen symbolic link and consequently be written immediately to storage media. [My guess is that with the symbolic link to /mnt/home being builtin, the contents of that mnt folder are also not copied to Storage unless you modify what other storage media to mount on bootup.]

When a Save is executed any other changes made during the current session will be written to the SaveFile/Folder. That includes the setting changes apparently made in /root/.config and others; the changes to documents you have in /root/my-documents and the web-page caches in /root/.cache or /root/spot/.cache, albeit the actual changes were only written to corresponding folders in /initrd/mnt/tmpfs/pup_rw. Which is why, as I said before, I run Puppies under Pupmode 13 with Automatic Save Removed; only install applications immediately on bootup; only Save what SFSes I want to have loaded immediately upon bootup and try to generally try not to have to make any changes except immediately upon bootup when there should not be anything in /initrd/mnt/tmpfs/pup_rw other than the changes I actually want to make.

mikesLr

[6502coder]

Can Linux actually update in-place only PART of a file on a storage medium?

Yes. This is how random-access databases work.

Interesting. I guess that would explain what I've noticed when doing backups of my savefiles: that the "last modified" date of a savefile never seems to change (it remains stuck at the creation date).

[Billtoo]

Xfce4 doesn't work in 32bit xenialpup-7.5 (works perfect in 64bit) so
I am giving radky's fbBox-5.0 a try as an alternative to JWM.
I'm still using gkrellm with fbBox because it's my favorite system
monitor.

It's working well, no problems so far.

[ethplorer]

if you have to do a frugal install on this, what are your menu.lst parameters inorder to boot the system?

i'm putting all the files in a directory named xenialce75 in the /mnt/sda1/
am i doing it correctly?

[foxpup]

if you have to do a frugal install on this, what are your menu.lst parameters inorder to boot the system?

i'm putting all the files in a directory named xenialce75 in the /mnt/sda1/
am i doing it correctly?

That's probably allright.
You give little info to know for sure. I suppose you are on an older machine with legacy boot? Is Grub4dos already installed? MBR and grldr and menu.lst? Do you dual boot with some Windows?
Anyway, have you run grub4dos just to generate a new menu.lst? It will probably find the new frugal install and make a new entry for it in menu.lst.

[ethplorer]

do you have updated chrome for this pup? tried installing deb files from ubuntu didn't work/ any help please

[mikeslr]

do you have updated chrome for this pup? tried installing deb files from ubuntu didn't work/ any help please

Puppies are "woofed"/built to be binary compatible with the distro-version whose binaries go into the building. That "binary-compatibility" makes it possible to install applications directly from the repos of the distro-version using PPM (Puppy Package Manager).

But, be wary about doing that. No Puppy is identical to the distro-version whose binaries were used. Packages built for the distro-version (in your case Xenial Xerus) are so built with the expectation that the foundation of Xenial Xerus will be present and --with specific reference to Xenial-Xerus' Chrome package-- that the system itself will operate in a particular manner.

Xenial-Xerus is a Multi-User system. Consequently its packages are designed to be installed into a User's unique directory. Puppy is a single-user system designed to run as "root"/Administrator. Usually that distinction doesn't matter.

But Gogle-Chrome [and its original and clones, Chromium, Iron, Slimjet and others] have been for over a year structured by Google NOT TO PERMIT IT TO BE RUN AS ROOT.

In order to run recent Chrome-and-clones under Puppies, they have to be configured to 'run-as-spot', a limited user. OscarTalks, peebee and others have undertaken the arduous tasks of packaging some-Chrome-clones in that manner. You'll find packages in the Additional Software Section. http://murga-linux.com/puppy/index.php?f=53.

There is no current version of Google-Chrome available. OscarTalks has a preference for Slimjet. You'll find a link to his latest version here, http://murga-linux.com/puppy/viewtopic. ... 08#1006708 and a link to peebee's slightly older version a couple posts above that.

I, personally, like OscarTalks' Iron 63, link from here: http://murga-linux.com/puppy/viewtopic. ... 05#1000505. While you're on that thread, read spiritwild's Nov 30, 2017 post about "user-agent". Some websites won't provide access unless you are using Chrome or Firefox, or they think you are.

Many still find Google-Chrome 48 (pre-Spot) adequate; link here: http://murga-linux.com/puppy/viewtopic. ... 954#909954. Also, pre-Spot is Iron 58. I still occasionally use it. You'll find a link on the Iron thread.

DO UNINSTALL the Chrome you obtained via PPM before trying any "chrome" browser. And if you decide on a "run as spot" browser, look for discussions about how using them effects your practice.

Lastly, in general, before installing anything via PPM that will access a Ubuntu repo, check the Additional Software Section to see if someone has published that application or its equivalent for your Puppy. It will save you a lot of frustration.

[roadkill13]

Hi ethplorer.

User mikesir has provided a quite a bit of good information in the above post. It is worth reading through. Since I am running Chrome (and have run Chromium) on XenialPup 64 I thought that I would give you a couple of quick tips.

If you installed the version of Chromium currently available in the puppy package manager it will run as user "spot". I am not sure why it won't run for you. philb666 pre-selected chromium-browser and google-chrome in the loginmanager applet to run as spot when installed on this version of puppy.

To confirm this, open a console and type "loginmanager" without the quotes. You will see a number of browsers listed, several of which have been pre-selected for use as user "spot".

Try running the browser from the console and see if it prints a message.

Also Mike Walsh provides a package for Google Chrome 64 which he updates periodically. It is pre-configured to run as "spot" and provides an additional utility for managing files in the /root/spot/downloads folder.

EDIT: I wasn't paying attention to the fact that you are using the 32 bit version of XenialPup or I would not have recommended the 64 bit package for Google Chrome. Obviously Google Chrome won't run on the 32 bit version of Xenialpup because Chrome only comes in the 64bit flavor on linux.

I have run Slimjet and Chromium on both setups with equal success on the 32bit version of XenialPup and on musher0's version of xenialpup. I apologize for my oversight.

[mikeslr]

Hi ethplorer,

I'm not familiar with Xenialpup 7.5, never having upgraded from their precedessors, 7.0.8.6, in either the 32-bit or the 64-bit versions. So, I may be missing out on quickpet and changes in PPM that roadkill13 discussed.

[ethplorer]

run as spot didn't work. yes i was using the 32bit. but thanks i now know there is a user spot. the program is installed but it kept prompting "Aw Snap" error.

i kept trying the ppm but no results still even after installing all the dependencies it asked. until i accidentally load the google_chrome-48.0.2564.116-i386.sfs in the /mnt/home ( which is from slacko) i wonder why it worked. its old version though. anyway i learn a lot. thanks

[thinkpadfreak]

ethplorer wrote:

> run as spot didn't work.

I run as spot the version of chromium which I installed from ubuntu deb packages.

To enable user spot, I open Login and Security Manager, uncheck "chromium-browser" and press "OK."
Then again I open Login and Security Manager, check "chromium-browser" and press "OK."

It is my understanding that user spot is enabled when the status is changed from "unchecked" (disabled) to "checked" (enabled).

[Bird Dog]

Hi ethplorer have you tried Palemoon. If you have a minimum P4 computer it receives updates, is the default browser for Xenialpup and I haven't had any problems (crashes etc)except when Palemoon switched to version 28 which quickpet will take care of when you update it.
Just want to thank some of the posters here for mentioning quickpet (was included in previous pups) and for the people who spend the time to fix problems which helps puppy users keep using this great operating system.

Thanks
Bird Dog

[glene77is]

BigPup

I have noticed an extra 30 odd seconds of System Boot Time
when switching from Tahr-pup to Xenial-pup.
IMO, the difference is do partly to the different software
used to control Disk Writes.
There is some difference in required Bootup time
as I test between several vintage PC's and a vintage LapTop,
and then between different types of bootable USB FlashDrives.
The same events occur on my bootable HardDrive system,
and my bootable FlashDrive systems.

Thus far:
(1) The Linux Xenial pup system seems to have a 30 second onger boot time
than the Linux Tahr pup system.
(2) Also two different USB FlashDrive brands have different times.
(3) IMO, there are soft-ware differences which cause these differences.

Focusing on the Xenial pup slow shutdown on USB :
Format Fat32 , so that M$ XP can read, in an emergency.

I have been using USB (SanDisk) for 7 years, bootable.
Normally, a 32 GB, SanDisk. Lupu to Xenial systems.

The SanDisk write speed is about as advertised.
with no stalling or waiting.
(1) LED blinks when working .
(2) LED stops blinking and control returns to system.

When I tried using a different ( Lexar ) USB,
then the write speed on shutdown slowed 50% more.

Measurements are by timing the wait until actual full return
to system and user control is allowed.

IMO,
The SansDisk USB appears to flush the out-going data immediately
into the Ram storage areas on the USB chip ,
and stops LED blinking just a second before real control is returned
to the system.

The Lexar USB appears to buffer the out-going data ,
and then shuffle it out to the final area on the layered RAM chips.
The Lexar LED blinking continues for a dozen extra seconds
before real control is returned to the system.

When using 128 GB and 256 GB USB flash drives,
the difference in shutdown speed of Lexar USB is a minute longer.

IMO,
I think this slower Real-Time Write speed ( Lexar )
is due to different soft-ware used by Lexar.
If Lexar is using a quick loading buffer for common small writes,
(1) then re-writing out to the real RAM area,
(2) then the user will see a quicker apparent save of 1 MB data files,
and will not notice that Lexar is re-shuffling the data.
(3) Very Large Data Saves will seem to go faster,
but the persistent write will be much slower.

I found this Lexar overall speed slow and un-acceptable
when doing backups of 100 GB of data onto a USB drive.
Therefore, I have continued to use the SansDisk
( with slightly slower large file save speed )
because the large file backup is saved and control returned to system
in a faster time. Control is returned to the system as expected.

For example of this problem
with Lexar backup on 100 GB data :
(1) I do a "save2flash" , and then request to "unmount' Lexar.
(2) then Xenial system stalls
(3) then Lexar blinking LED continues for several minutes.
(4) then Lexar LED stops blinking,
(5) then Xenial system is responsive again.

*******
I was curious about where Xenial was spending much of its time
during sysinit , so I wrote logging-with-date-time routines
into all the scripts running : rc.sysinit, sfs_load, fixmenus, init.
Results are: sfsload is the script using most of the time, not fixmenus.

So, that is for today.

PS: Your Results May Vary.


.

[sendwho]

I used the use the full drive option and now the pupsave backup doesn't do anything when I try to use it. How can I take backup of my system, in order to restore it later?

[bigpup]

Pupsave Backup program only works for making backups of the save file/folder used in frugal installs.

What exact program are you trying to use?