Page 13 of 15

Posted: Wed 27 Feb 2013, 01:55
by zygo
Thanks rerwin,

Puppy *can* fully change between my 2 network IFs.

I've replied on the Frisbee thread http://murga-linux.com/puppy/viewtopic. ... &start=180

Thanks gcmartin,

I have a tower PC but prefer "pay as you go" ISPs. Mr reply on the Frisbee thread should answer your questions.

Another frisbee version to avoid endless wpa log

Posted: Wed 27 Feb 2013, 23:55
by rerwin
I have noticed that the /tmp/wpa_supplicant.log file gets updated with many lines every time the networks are scanned, every 30 seconds or maybe more frequently. This is unacceptable for normal operation for long periods.

To correct this excessive logging, I added a control checkbox for when debug-level logging is needed, but otherwise to be left unchecked (disabled). The updated frisbee-1.0 package, 20130228 (was 20130227), is at the usual place:
http://murga-linux.com/puppy/viewtopic. ... 092#684092

Please test this version soon, so it can be considered for inclusion in the release version of precise 5.5. Thanks.
Richard

Precise-5.4.93

Posted: Sat 02 Mar 2013, 01:04
by sszindian
Starting to get a little confusing as to where to post results for Precise with all the various threads, and new versions some things starting in the middle of old threads, hope this gets to the eyes and ears of BK

Precise-5-4-93

Did an 'upgrade' to Precise-5.4.91 (was going to upgrade to 5.4.92 but BK beat me to the draw with 5.4.93 just after downloading 5.4.92... boy, when he's on a roll :)

Anyway, upgrade went fine, usual loss of personal apps & icons, no big deal.

All programs appear to be working as before with one little 'quirk' as in previous versions... analog clock disappears from pwidgets ( 2.4.0 & 2.4.1) on desktop every now and then? Appears to me it isn't a pwidgets fault as this also happens in Dpup-Precise, Dpup-Wheezy and Slacko... Hmmm... maybe a big bad wolf thingy? Just guesing of course?

Previous version of 5.4.91 was 'dropping WiFi' a few times per day, only re-coop was a cold hard button shutdown... have to see what happens for a few days with 5.4.93? (Using: Linksys AE1000 usb plugin... Verizon Router - built into DSL modem- and rt2800usb setup - which was automatic.) Hope it's now fixed! There are some other distro's that have an auto re-connect after a drop, that would be a nice feature to consider for somewhere down the road?

OK... 2-hrs after writing the above on the WiFi thing... WiFi DROPPED, again a cold shutdown. The WiFi hung while my wife was posting a message in Facebook. After thinking on this, I decided to try to remove Pwidgets as I have a Weather module installed that keeps reaching out for updates, I thought this may be the problem with the drops. Going to the PPM I REMOVED Pwidgets (PPM said it in fact removed Pwidgets). Well, Pwidgets was still on the desktop! A reboot and although Pwidgets was said to be removed, all the modules still appear on the desktop? I re-installed Pwidgets and just removed the weather module (that took) so now I maybe can see if that has anything to do with the WiFi drops? Don't understand why Pwidgets DID NOT remove after being un-installed through the PPM???

EDIT: It's not the Pwidgets weather module, another drop without it!!!!

Seamonkey 15 preforming well!

Youtube viewing... some of the fonts are showing up as 'PCOMPASS FONTS' (squiggly little arrowheads) instead of regular text? (That's a new one for me.)

OK... that's what I have so far... testing continues!

>>>---Indian------>

Posted: Sat 02 Mar 2013, 17:51
by mchabez
A few more issues:

1. Glipper tray icon's background is white, not transparent.

2. Package deps like "upstart-job" aren't resolved in PPM. There is no such package name "upstart-job" in the Ubuntu Precise repos, as it's a virtual package provided by "upstart". One such package dependent on upstart-job is "binfmt-support".

Posted: Sat 02 Mar 2013, 19:55
by artsown
Barry, 01Mico requested that I post this on one of your threads. It
concerns a bug in the startup script where optical media are being
interrogated. A freeze occurs forcing a power down. I had learned to
live with this happening once in awhile but for some reason his latest
Slacko (5.4.0.5) is really bad in this regard (10 freezes in a row). So
I finally opened the case and disconnected the cables from the DVD
Drive. That allowed Slacko to startup.

The drive is a fairly new LiteOn Super All Write DVD Writer. It's been
causing this trouble going back to your Precise 5,4.3 (or so). It doesn't
happen if I manage to start up and create a Save file. It only occurs
when trying to start a pup the first time.

Art

/tmp permissions hazard - fixed

Posted: Sat 02 Mar 2013, 23:00
by rerwin
01micko has been after me to clean up the /tmp directory permisions in my frisbee-1.0 packages, because it created situtaion that jeopardizes puppy stablity/rebustness. I have corrected that problem in the 20130228 version now available in the usual places.

Here is his patient explanation to me:
Here is the problem.. I wish i could find the link.

If permissions on /tmp are not drwxrwxrwt (or 1777) some cups drivers do not work. We found this problem occurring time and time again in Lupu, and it was mainly because of bad DEB packages with 755 or drwxr-xr-x. A package that contains those permissions will change the permissions on /tmp on the system. Some debs were so bad that they weren't even owned by root! They would default to users:500.

I know you delete the file directly but the damage is done. As soon as the pet gets extracted to it's path the permissions change, check your /tmp, permissions will likely be wrong.

It also affects users who want to use the universal installer to install to a USB drive, I don't know why but it is reported in the slacko-5.4 bug thread, I can find that link.. http://murga-linux.com/puppy/viewtopic. ... 630#677630

I hope this gives you some understanding. I had problems with zigbert's Pmusic too, as he was using /tmp as a temporary location to install other stuff, rox right clicks I think, that don't work in woof.

It's not for me, because woof changes the permissions to 1777 when the build is almost complete. it's for the users out there installing and testing. They may find later on that printing has failed or they can't get their USB stick to work. These problems are difficult to debug. If I see a post like that I'll ask for the output of ls -l /

NOTE: You only need to change the permissions on /tmp and no other directory in your package tree.
. . .
In rox, if you select "properties" you get the checkboxes, all on the left (9 of them) should be checked and on the right the bottom one.. "sticky" is the label.
The link points to this conversation:
Master_wrong
. . .
during install into usb flashdisk there is syslinux error
Quote:
possibly unsafe /tmp/ permissions

fix
Quote:
chmod 1777 /tmp

then run installer
_________________

01micko

Perms are already 1777 on /tmp. You may have installed a bad package.
_________________

SFR

@Master_wrong & @01micko

That's strange, but in my case /tmp sometimes has 1777 and sometimes 777.
Eg. in my install (pmedia=ataflash on internal SATA HDD) it had 777 by default and I couldn't make it to preserve 1777 after reboot.
(/initrd/pup_ro2/tmp has 777 as well, perhaps that's why?)
I simply added chmod 1777 /tmp to rc.sysinit to achieve this.

On the other hand, when I switch to pmedia=atahd, /tmp suddenly becomes 1777 and, in addition, it appears as a mount point.
(I checked this a few minutes ago in VBox, 5.4-PAE)

Also, pristine boot without a savefile also gives /tmp 777.

IIRC I noticed this also in 5.3.3, but I didn't report it, because usually I have "luck" of experiencing "it's_just_me" kind of issues...
Attached are small mods to dir2pet and installpkg.sh to ensure correct permissions in a .pet and after installation of any package. Here are the difference listings:

Code: Select all

--- dir2pet-5492	2013-01-13 19:09:21.000000000 -0500
+++ dir2pet	2013-03-02 16:33:48.000000000 -0500
@@ -13,6 +13,7 @@
 #w482 if preexisting pet.specs, read fields from it.
 #100201 improve reading of pet.specs
 #100303 fix bug detection arch-dependent pkg.
+#130302 ensure package tmp directory, if any, has all permissions.
 
 if [ ! $1 ];then
  echo "It is required to invoke this script like this:
@@ -347,6 +348,8 @@
  exit
 fi
 
+[ -d $DIRPKG/$BASEPKG/tmp ] && chmod 1777 $DIRPKG/$BASEPKG/tmp #130302
+
 cat /tmp/petspec_db_entry | tail -n 1 > $DIRPKG/$BASEPKG/pet.specs
 
 echo



--- installpkg-5492.sh	2013-02-19 08:35:15.000000000 -0500
+++ installpkg.sh	2013-03-02 16:31:22.000000000 -0500
@@ -47,6 +47,7 @@
 #130114 revert 130112 "multiarch symlinks now optional".
 #130126 'categories.dat' format changed.
 #130219 grep, ignore case.
+#130302 ensure tmp directory has all permissions after package expansion.
 
 #information from 'labrador', to expand a .pet directly to '/':
 #NAME="a52dec-0.7.4"
@@ -358,6 +359,8 @@
 mv /*.xpm /usr/local/lib/X11/mini-icons/ 2>/dev/null
 mv /*.png /usr/local/lib/X11/mini-icons/ 2>/dev/null
 
+ls -dl /tmp | grep -q '^drwxrwxrwt' || chmod 1777 /tmp #130302
+
 #post-install script?...
 if [ -f /pinstall.sh ];then #pet pkgs.
  chmod +x /pinstall.sh

delayed run

Posted: Sun 03 Mar 2013, 03:51
by phdzaps
I am not sure if it is a real improvement or not but I think it is:
in /usr/sbin/delayedrun, I change line 199 to read:
[ -x "$a" ] && exec $a &
Things seem to run slightly better for me after. This is in multiple versions of puppy also.

USB flash memory stick not listed in pmount

Posted: Sun 03 Mar 2013, 12:33
by zygo
The following USB flash memory stick is not listed in pmount in PP5.4.92 but was in the previous version of PP.

Code: Select all

Manufacturer=SanDisk
Product=Cruzer
VendorID=0781  ProductID=5530  KERNEL-DRIVER(builtin)=usb-storage
I have one other which I've been using for years and that works fine in both versions of PP. I've been using the new one for only a few months. It's light displays the usual flashing while if boots and when it's ready.

Edit: It's no longer listed in pmount in Quirky 130. I suppose it died.

Posted: Sun 03 Mar 2013, 21:40
by futwerk
some new backgrounds.

Re: delayed run

Posted: Mon 04 Mar 2013, 10:05
by BarryK
phdzaps wrote:I am not sure if it is a real improvement or not but I think it is:
in /usr/sbin/delayedrun, I change line 199 to read:
[ -x "$a" ] && exec $a &
Things seem to run slightly better for me after. This is in multiple versions of puppy also.
Errr, that's what it already is.

Unload SFS

Posted: Mon 04 Mar 2013, 20:52
by josepinto
Hi,

I cannot unload SFS using SFS load on the fly.

José Pinto

Posted: Mon 04 Mar 2013, 20:54
by 666philb
updated quickpet_precise
added a load of games http://www.murga-linux.com/puppy/viewtopic.php?t=83642

Posted: Mon 04 Mar 2013, 21:17
by anikin
An attempt to remaster precise-4.5.93 (no CD drive, but the ISO is mounted) ends with this error warning - see attached image. The one-line fix suggested in wary/racy thread doesn't help. Any ideas?

PS: this was initially posted in a wrong thread, sorry.

edit: a quick test - remaster doesn't stumble on this ISO in dpup exprimo-3.6.2. Something must have changed in the script. See another attached image.

Posted: Mon 04 Mar 2013, 22:27
by 666philb
hi anikin

did you see my reply in the other thread?

Posted: Mon 04 Mar 2013, 22:52
by anikin
666philb wrote:hi anikin

did you see my reply in the other thread?
hi,
the iso is mounted, nonetheless the script doesn't work for me. This puppy runs off of an ext4 sd card. Can it be that ext4 isn't fully supported? I can see strange boot messages saying fs is ext2 instead of ext4?

Posted: Mon 04 Mar 2013, 23:31
by kurtdriver
Probably more of a feature request than a bug, but when I attempted to install puppy to a desktop computer this afternoon, the universal installer wouldn't give me the option of a full installation, it assumed I wanted to keep Windows. I had to run mkfs.ext4 on them first. A newcomer to Linux wouldn't have known to do that.

Yet another frisbee update, 20130304

Posted: Tue 05 Mar 2013, 00:06
by rerwin
Thanks to mavrothal and peebee keeping after me, I think I have solved the issue of frisbee not running correctly in racy and wary, and now also in precise. Frisbee, all along, has used the wpa_supplicant "-f" option to specify a "debug" log file. That is apparently configurable when compiling wpa_supplicant. Now, it is rejected, preventing wpa_supplicant from starting. Since the same output goes to stdout, the fix simply redirects the output from the startup of wpa_supplicant, to the log file. I feel it is important that this fix be included in the next beta release, assuming the testing of it goes well.

Barry, please hold off until we get confirmation that my fix works. For anyone who is willing to test it, the package is here:
http://murga-linux.com/puppy/viewtopic. ... 092#684092

EDIT early 3/5/2013: I have verified that the 20130304 frisbee package restores frisbee's normal functionality with wireless networks, on precise 5.4.92. Mavrothal is testing, too, probably on Racy 5.5.
Richard

Re: /tmp permissions hazard - fixed

Posted: Tue 05 Mar 2013, 00:20
by BarryK
rerwin wrote:01micko has been after me to clean up the /tmp directory permisions in my frisbee-1.0 packages, because it created situtaion that jeopardizes puppy stablity/rebustness. I have corrected that problem in the 20130228 version now available in the usual places.

Here is his patient explanation to me:
Here is the problem.. I wish i could find the link.

If permissions on /tmp are not drwxrwxrwt (or 1777) some cups drivers do not work. We found this problem occurring time and time again in Lupu, and it was mainly because of bad DEB packages with 755 or drwxr-xr-x. A package that contains those permissions will change the permissions on /tmp on the system. Some debs were so bad that they weren't even owned by root! They would default to users:500.

I know you delete the file directly but the damage is done. As soon as the pet gets extracted to it's path the permissions change, check your /tmp, permissions will likely be wrong.

It also affects users who want to use the universal installer to install to a USB drive, I don't know why but it is reported in the slacko-5.4 bug thread, I can find that link.. http://murga-linux.com/puppy/viewtopic. ... 630#677630

I hope this gives you some understanding. I had problems with zigbert's Pmusic too, as he was using /tmp as a temporary location to install other stuff, rox right clicks I think, that don't work in woof.

It's not for me, because woof changes the permissions to 1777 when the build is almost complete. it's for the users out there installing and testing. They may find later on that printing has failed or they can't get their USB stick to work. These problems are difficult to debug. If I see a post like that I'll ask for the output of ls -l /

NOTE: You only need to change the permissions on /tmp and no other directory in your package tree.
. . .
In rox, if you select "properties" you get the checkboxes, all on the left (9 of them) should be checked and on the right the bottom one.. "sticky" is the label.
The link points to this conversation:
Master_wrong
. . .
during install into usb flashdisk there is syslinux error
Quote:
possibly unsafe /tmp/ permissions

fix
Quote:
chmod 1777 /tmp

then run installer
_________________

01micko

Perms are already 1777 on /tmp. You may have installed a bad package.
_________________

SFR

@Master_wrong & @01micko

That's strange, but in my case /tmp sometimes has 1777 and sometimes 777.
Eg. in my install (pmedia=ataflash on internal SATA HDD) it had 777 by default and I couldn't make it to preserve 1777 after reboot.
(/initrd/pup_ro2/tmp has 777 as well, perhaps that's why?)
I simply added chmod 1777 /tmp to rc.sysinit to achieve this.

On the other hand, when I switch to pmedia=atahd, /tmp suddenly becomes 1777 and, in addition, it appears as a mount point.
(I checked this a few minutes ago in VBox, 5.4-PAE)

Also, pristine boot without a savefile also gives /tmp 777.

IIRC I noticed this also in 5.3.3, but I didn't report it, because usually I have "luck" of experiencing "it's_just_me" kind of issues...
Attached are small mods to dir2pet and installpkg.sh to ensure correct permissions in a .pet and after installation of any package. Here are the difference listings:

Code: Select all

--- dir2pet-5492	2013-01-13 19:09:21.000000000 -0500
+++ dir2pet	2013-03-02 16:33:48.000000000 -0500
@@ -13,6 +13,7 @@
 #w482 if preexisting pet.specs, read fields from it.
 #100201 improve reading of pet.specs
 #100303 fix bug detection arch-dependent pkg.
+#130302 ensure package tmp directory, if any, has all permissions.
 
 if [ ! $1 ];then
  echo "It is required to invoke this script like this:
@@ -347,6 +348,8 @@
  exit
 fi
 
+[ -d $DIRPKG/$BASEPKG/tmp ] && chmod 1777 $DIRPKG/$BASEPKG/tmp #130302
+
 cat /tmp/petspec_db_entry | tail -n 1 > $DIRPKG/$BASEPKG/pet.specs
 
 echo



--- installpkg-5492.sh	2013-02-19 08:35:15.000000000 -0500
+++ installpkg.sh	2013-03-02 16:31:22.000000000 -0500
@@ -47,6 +47,7 @@
 #130114 revert 130112 "multiarch symlinks now optional".
 #130126 'categories.dat' format changed.
 #130219 grep, ignore case.
+#130302 ensure tmp directory has all permissions after package expansion.
 
 #information from 'labrador', to expand a .pet directly to '/':
 #NAME="a52dec-0.7.4"
@@ -358,6 +359,8 @@
 mv /*.xpm /usr/local/lib/X11/mini-icons/ 2>/dev/null
 mv /*.png /usr/local/lib/X11/mini-icons/ 2>/dev/null
 
+ls -dl /tmp | grep -q '^drwxrwxrwt' || chmod 1777 /tmp #130302
+
 #post-install script?...
 if [ -f /pinstall.sh ];then #pet pkgs.
  chmod +x /pinstall.sh
Thanks, done in Woof.

Posted: Tue 05 Mar 2013, 00:31
by BarryK
mchabez wrote:2. Package deps like "upstart-job" aren't resolved in PPM. There is no such package name "upstart-job" in the Ubuntu Precise repos, as it's a virtual package provided by "upstart". One such package dependent on upstart-job is "binfmt-support".
Yes, I see. OK, I will have a go at fixing that.

Posted: Tue 05 Mar 2013, 06:28
by Ray MK
 Precise Puppy version 5.4.93, released Feb 2013

Linux puppypc17519 3.8.0 #1 SMP Wed Feb 20 17:29:16 GMT-8 2013 i686 i686 i386 GNU/Linux

# free
total used free shared buffers
Mem: 366580 351240 15340 0 13320
-/+ buffers: 337920 28660
Swap: 658660 35992 622668

#top
Mem: 353560K used, 13020K free, 0K shrd, 15580K buff, 235568K cached
CPU: 15% usr 5% sys 0% nic 79% idle 0% io 0% irq 0% sirq
Load average: 0.37 0.61 0.50 1/94 25031

This running extremely well on my 10yr old Acer laptop. Usual setup - manual frugal to ext3 - booting with grub4dos.

Resources are only fractionally more than Wary 5.5 and much less than Racy 5.5 - temp seems similar to Wary - quite remarkable - and with k3.8.0.

Report whilst running SM browser, Abiword and top.

This is one stunning Puppy - very snappy (even on 10yr old kit) - Luvly.