Re: objprelink
On Sat, Sep 15, 2001 at 07:57:12PM -0400, Jason Boxman wrote: > On Saturday 15 September 2001 06:03 pm, Ivan E. Moore II wrote: > > On Sat, Sep 15, 2001 at 11:07:29PM +0200, Maximilian Reiss wrote: > > > qt-x11 (2:2.3.1-13) unstable; urgency=low > > > . > > >* Due to better sense don't use objprelink..wait for the proper > > > prelinker stuff to show up > > > > > > Does this means it is compiled _without_ objprelink? > > > > Well I would htink that "don't use objprelink" would me don't use it. :) > > D'oh. I compiled the whole slew of KDE 2.2.0 Sid sourc packages on Woody > with objprelink since stuff was blowing up without it. I compiled a Sid > binutils and everything, too. Was that bad? objprelink built packages have not seen any problems that I know of directly. Ivan -- Ivan E. Moore II [EMAIL PROTECTED] http://snowcrash.tdyc.com GPG KeyID=90BCE0DD GPG Fingerprint=F2FC 69FD 0DA0 4FB8 225E 27B6 7645 8141 90BC E0DD
Re: objprelink and lintian errors
On Sat, Sep 15, 2001 at 05:54:03PM -0500, Ben Burton wrote: > > > yes...that's a side effect of using objprelink. > > So does this mean package maintainers should stop using objprelink because > it's causing policy violations? Or is the lintian check slightly broader > than policy and objprelink within the margin of error? it's not a policy violation..I think it's just a side affect and a false positive. the only reason package maintainers should stop using objprelink is becuase it's a hack and not the correct solution. Ivan -- Ivan E. Moore II [EMAIL PROTECTED] http://snowcrash.tdyc.com GPG KeyID=90BCE0DD GPG Fingerprint=F2FC 69FD 0DA0 4FB8 225E 27B6 7645 8141 90BC E0DD
Re: objprelink and lintian errors
Am Sonntag, 16. September 2001 07:59 schrieb Ivan E. Moore II: > On Sat, Sep 15, 2001 at 05:54:03PM -0500, Ben Burton wrote: > > > yes...that's a side effect of using objprelink. > > > > So does this mean package maintainers should stop using objprelink > > because it's causing policy violations? Or is the lintian check slightly > > broader than policy and objprelink within the margin of error? > > it's not a policy violation..I think it's just a side affect and a false > positive. > > the only reason package maintainers should stop using objprelink is becuase > it's a hack and not the correct solution. > > Ivan With now having kde starting MUCH slower. Great work. Max
kde an objprelink
Is somebody up to build kde WITH objprelink from now on? Sadly I do not have the webspace where I could supply them as external apt source. I also use KDE on computers slower then 300 Mhz and would really like to have kde start in 10 seconds on them, not 43 seconds. ;-) Max
Re: objprelink and lintian errors
On Sunday 16 September 2001 13:39, Maximilian Reiss wrote: > > > > yes...that's a side effect of using objprelink. > > > > > > So does this mean package maintainers should stop using objprelink > > > because it's causing policy violations? Or is the lintian check > > > slightly broader than policy and objprelink within the margin of error? > > > > it's not a policy violation..I think it's just a side affect and a false > > positive. > > > > the only reason package maintainers should stop using objprelink is > > becuase it's a hack and not the correct solution. > > > > Ivan > > With now having kde starting MUCH slower. Great work. Yeah, that sucks. What's wrong with using the hack until the correct solution is available? Even the KDE folks recommend this. Ciao...
Re: objprelink and lintian errors
Am Sonntag, 16. September 2001 13:46 schrieb Stephan Jaensch: > On Sunday 16 September 2001 13:39, Maximilian Reiss wrote: > > > > > yes...that's a side effect of using objprelink. > > > > > > > > So does this mean package maintainers should stop using objprelink > > > > because it's causing policy violations? Or is the lintian check > > > > slightly broader than policy and objprelink within the margin of > > > > error? > > > > > > it's not a policy violation..I think it's just a side affect and a > > > false positive. > > > > > > the only reason package maintainers should stop using objprelink is > > > becuase it's a hack and not the correct solution. > > > > > > Ivan > > > > With now having kde starting MUCH slower. Great work. > > Yeah, that sucks. What's wrong with using the hack until the correct > solution is available? Even the KDE folks recommend this. > Might be a problem to get KDE 2.2(.1) in woody. I don't know. Max
Re: objprelink and lintian errors
Please, please, read the netiquette, and stop including the whole message when replying. My mousewheel is almost broken because of this :) To be on-topic : I didn't noticed any difference betweeen packages created with and without objprelink. Hmm, it is maybe my computer's fault? -- Putz Ákos Email : [EMAIL PROTECTED]
Re: kde an objprelink
On Sun, Sep 16, 2001 at 01:46:50PM +0200, Maximilian Reiss wrote: > Is somebody up to build kde WITH objprelink from now on? > Sadly I do not have the webspace where I could supply them > as external apt source. > > I also use KDE on computers slower then 300 Mhz and would really > like to have kde start in 10 seconds on them, not 43 seconds. ;-) this was sent off to me..and I have spent time consulting with our binutils maintainer as well as doing tests of my own. --- "On Sun, Sep 09, 2001 at 11:57:32AM -0400, Jack Howarth wrote: > Jakub, >HJ told me that your work on the prelinker > set to eventually be adopted into Linux. I ask > because debian has started to use objprelink > for now. If you ask me, I think it is a bad decision. objprelink means slightly slower execution time even where it really doesn't buy anything. Also, at least in the version which is floating around e.g. the ia-32 stub is too long and inefficient. Also, with objprelink they have to muck with all packages. If debian wants good startup speeds, then I think the no1 thing is to use recent binutils (and on arches where it works enable -z combreloc by default). Together with current glibc, already doing just this speeds startup of large KDE programs a lot, the symbol lookup is the most expensive thing (together with unnecessary COW) and glibc's lookup cache together with -z combreloc hits a lot: LD_DEBUG=statistics kmail 14751: 14751: runtime linker statistics: 14751:total startup time in dynamic loader: 226875015 clock cycles 14751: time needed for relocation: 223635459 clock cycles (98.5%) 14751: number of relocations: 24432 14751:number of relocations from cache: 45710 14751: time needed to load objects: 2969083 clock cycles (1.3%) Also, these binutils already reserve .dynamic entries for prelink, so rebuilding everything with this is really worthy." --- Jakub Jelinek Debian's binutils already usine -z combreloc (or will with the next release acording to the binutils maintainer). glibc's changes will be taken care of with the next major release of glibc (after woody's release). And to set something straight, KDE does not recommend using objprelink. Some folks within KDE do, however objprelink has caused problems in certain situations. Ivan -- Ivan E. Moore II [EMAIL PROTECTED] http://snowcrash.tdyc.com GPG KeyID=90BCE0DD GPG Fingerprint=F2FC 69FD 0DA0 4FB8 225E 27B6 7645 8141 90BC E0DD
Re: objprelink and lintian errors
On Sun, Sep 16, 2001 at 01:39:23PM +0200, Maximilian Reiss wrote: > Am Sonntag, 16. September 2001 07:59 schrieb Ivan E. Moore II: > > On Sat, Sep 15, 2001 at 05:54:03PM -0500, Ben Burton wrote: > > > > yes...that's a side effect of using objprelink. > > > > > > So does this mean package maintainers should stop using objprelink > > > because it's causing policy violations? Or is the lintian check slightly > > > broader than policy and objprelink within the margin of error? > > > > it's not a policy violation..I think it's just a side affect and a false > > positive. > > > > the only reason package maintainers should stop using objprelink is becuase > > it's a hack and not the correct solution. > > > > Ivan > > With now having kde starting MUCH slower. Great work. What an asshole you are I swear. I work my ass off to provide the proper and most stable/secure KDE environment for Debian and you start being a prick. Screw you! You should first go off and learn some manners, then go off and learn some programming skills so that you would first learn what your talking about before you start being a dick. Damn, this is why I'm spending more time away from the computer these days. Ivan -- Ivan E. Moore II [EMAIL PROTECTED] http://snowcrash.tdyc.com GPG KeyID=90BCE0DD GPG Fingerprint=F2FC 69FD 0DA0 4FB8 225E 27B6 7645 8141 90BC E0DD
Re: objprelink and lintian errors
On Sun, Sep 16, 2001 at 10:15:45AM -0600, Ivan E. Moore II wrote: > On Sun, Sep 16, 2001 at 01:39:23PM +0200, Maximilian Reiss wrote: > > Am Sonntag, 16. September 2001 07:59 schrieb Ivan E. Moore II: > > > On Sat, Sep 15, 2001 at 05:54:03PM -0500, Ben Burton wrote: > > > > > yes...that's a side effect of using objprelink. > > > > > > > > So does this mean package maintainers should stop using objprelink > > > > because it's causing policy violations? Or is the lintian check > > > > slightly > > > > broader than policy and objprelink within the margin of error? > > > > > > it's not a policy violation..I think it's just a side affect and a false > > > positive. > > > > > > the only reason package maintainers should stop using objprelink is > > > becuase > > > it's a hack and not the correct solution. > > > > > > Ivan > > > > With now having kde starting MUCH slower. Great work. > > What an asshole you are I swear. I work my ass off to provide the proper > and most stable/secure KDE environment for Debian and you start being a prick. > Screw you! You should first go off and learn some manners, then go off and > learn some programming skills so that you would first learn what your talking > about before you start being a dick. > > Damn, this is why I'm spending more time away from the computer these days. ohh yea..and to boot... the *only* package that is currently not being built with objprelink out of the ones that were is libqt...which means even if it were speeding things up, you would have lost jack shit in speed as most of the speed up comes from the KDE packages themselves...I guess libqt is ultra cool and ignores that and just grindes everything to a halt anyways. -- Ivan E. Moore II [EMAIL PROTECTED] http://snowcrash.tdyc.com GPG KeyID=90BCE0DD GPG Fingerprint=F2FC 69FD 0DA0 4FB8 225E 27B6 7645 8141 90BC E0DD
Re: kde an objprelink
> "On Sun, Sep 09, 2001 at 11:57:32AM -0400, Jack Howarth wrote: > > Jakub, > >HJ told me that your work on the prelinker > > set to eventually be adopted into Linux. I ask > > because debian has started to use objprelink > > for now. > > If you ask me, I think it is a bad decision. > objprelink means slightly slower execution time even where it really > doesn't buy anything. Never seen in reality. >Also, at least in the version which is floating > around e.g. the ia-32 stub is too long and inefficient. Thats right > Also, with objprelink they have to muck with all packages. > > If debian wants good startup speeds, then I think the no1 thing is to use > recent binutils (and on arches where it works enable -z combreloc by > default). > Together with current glibc, already doing just this speeds startup of > large KDE programs a lot, the symbol lookup is the most expensive thing > (together with unnecessary COW) and glibc's lookup cache together with -z > combreloc hits a lot: The versions in sid? > > LD_DEBUG=statistics kmail > 14751: > 14751: runtime linker statistics: > 14751:total startup time in dynamic loader: 226875015 clock cycles > 14751: time needed for relocation: 223635459 clock cycles > (98.5%) 14751: number of relocations: 24432 > 14751:number of relocations from cache: 45710 > 14751: time needed to load objects: 2969083 clock cycles (1.3%) > > Also, these binutils already reserve .dynamic entries for prelink, so > rebuilding everything with this is really worthy." >
SCSI (aic7xxx) and KDE
Ivan,
This may end up as 'off topic' as the worst thread so far. The
situtaion is:
I was given a fantastic (early) Christmas present, and it's causing me more
gray hairs that I appreciate.
Is KDE involved, yes.
Is KDE at fault, I'm in doubt
This mail is going to be long, drawnout, and contain a shitpot of info. I
don't know how else I can get it all included. (shrug)
SYSTEM INFO:
Tyan ThunderBolt motherboard with current BIOS upgrade
Dual P3 750 cpu's
Onboard Adaptec AIC-7896 scsi bios V2.11
Western Digital WDE ULTRA3 ULTRA2-LVD (sda)
TDK CD-RW (hda & scd0)
Creative 8x DvD (hdb & scd1)
Kenwood True-X CD-ROM (hdc & scd2)
{The above devices are all seen as SCSI devices due to the SCSI
emulation
required for the CD-RW}
IOMega internal 100 Zip (hdd)
{IDE FLOPPY support is enabled in the kernel to support the Zip}
Standard floppy
GForce2 MX 32 meg video card
512 meg RAM (256 x 2)
(Ch A, SCSI ID: 0 WDIGTLWDE18310 ULTRA3 ULTRA2-LVD - Hard
Disk 0)
ERROR MESSAGE (captured, by hand)
scsi : aborting command due to timeout : pid 30523, scsi0, channel 0, id 0,
lun 0 Read (10) 00 01 ab d1 9d 00 00 08 00
SCSI hist 0 abort (pid 30522) timed out - resetting
SCSI bus is being reset for host 0 channel 0.
To date, the scsi bus has never reset correctly.
Upon startup of KDE2.x, the 'initialization' screen will freeze at the
point
of 'INITIALIZING PERIFERALS'. The hard drive light will come on and stay on,
the machine will stop responding. A full power off/power on is required to
reboot. The machine will boot, post a message that the partition is corrupted
and ask me to either use CTRL D or roots' pass for maintanence. Running
e2fsck -b 8192 (or 8193) just gives me a repeat of the error stating it can't
find one of the clusters (? not the correct term, damnit!)
After reading thru the /usr/src/linux/drivers/scsi/README.aic7xxx file,
I
added an 'append="aic7xxx=panic_on_abort"' to the lilo.conf which helped, in
that I did NOT have a corruptd partition until after the fifth reboot.
I have successfully ran (for some time) BlackBox & FVWM without any
errors
or complaints.
Today I finally was able to cause the same sort of behavior in
Enlightenment. It wasn't identical, but close. The drawback to this trouble
shooting is when I loose the partition, I LOOSE the partition. It's a full
reload. The only CD's I have to date are POTATO so I end up spending a number
of hours downloading to reinstall to the point where I can again begin
troubleshooting.
(Hendrik, yes I took yer advice and was going to build the CD's, but a
friend stopped by, and thinking he would assist, told dselect to erase the
downloaded files while I was on the phone to a client.}
I've gone thru a number of hours of google searches, and found an
enourmous
amount of data on the aic7xxx chipset, and some of the earlier problems they
had with this chipset. But nothing seems to apply to what I'm experiencing.
Would KDE be causing these failures so quickly due to the amount of hd
use
it has during 'startup'? I know we use a number of quick 'temp' files during
the startup procedure. (duder, i'm grasping at straws here) This machine run
Win2K (massive shudder) just fine. I've beat the hell outta it in Windon't
with a number of apps and have seti running on it now for six dayz while
defragmenting, scanning, moving, deleteing, etc and no errors.
I've used kernels from 2.2.17 up to 2.4.9.
I've enable smp, disabled smp, called it bad names, bribed it, etc. :--)
I've used POTATO, WOODY, and SID
Anyhow, you get the idea.
Does anyone have anything (urls, readme's, info, man pages, etc) that
might
help me shed some light on what I'm battling against?
--
__
OutCast Computer Consultants of Central Oregon
http://outcast-consultants.redmond.or.us
[EMAIL PROTECTED]
(541) 504-1388
Toll Free (866) 562-7160
Via IRC at; 205.227.115.251:6667:#OutCasts
Via ICQ: UIN 138930
"Failure is not an option...it's bundled with Microsoft"
-anonymous-
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
"Software is like sex. They're both better when they're free!!" - Linus
Torvalds
"As a computing professional, I believe it would be unethical for me to
advise, recommend, or support the use (save possibly for personal
amusement) of any product that is or depends on any Microsoft product."
Re: SCSI (aic7xxx) and KDE
On Sunday 16 September 2001 13:21, G. L. `Griz' Inabnit wrote: > Ivan, Hey! You help me get this running, you've got a dualie you can compile on!! How's that for an offer? :--) -- __ OutCast Computer Consultants of Central Oregon http://outcast-consultants.redmond.or.us [EMAIL PROTECTED] (541) 504-1388 Toll Free (866) 562-7160 Via IRC at; 205.227.115.251:6667:#OutCasts Via ICQ: UIN 138930 "Failure is not an option...it's bundled with Microsoft" -anonymous- Hi! I'm a .signature virus! Copy me into your ~/.signature, please! "Software is like sex. They're both better when they're free!!" - Linus Torvalds "As a computing professional, I believe it would be unethical for me to advise, recommend, or support the use (save possibly for personal amusement) of any product that is or depends on any Microsoft product."
koffice
Hi all, I put the line deb http://kde.debian.net potato main crypto optional kde2 in my sources.list, updated and wanted to upgrade to the new koffice. But apt-get is telling me: >apt-get -d install kchart Reading Package Lists... Done Building Dependency Tree... Done Sorry, kchart is already the newest version 0 packages upgraded, 0 newly installed, 0 to remove and 0 not upgraded. > Installed version is: > dpkg -s kchart Package: kchart Status: install ok installed Priority: optional Section: x11 Installed-Size: 1024 Maintainer: Ben Burton <[EMAIL PROTECTED]> Source: koffice Version: 1:1.1-beta1-0.potato2 Replaces: koffice-common Depends: kdelibs3 (>= 4:2.1.2-0), libc6 (>= 2.1.2), libjpeg62, liblcms (>= 1.07-0), libmng1 (>= 1.0.0-0), libpng2, libqt2 (>= 2:2.3.0-final-0) | libqt2-gl (>= 2:2.3.0-final-0), libstdc++2.10, libz1, xlib6g (>= 3.3.6-4) Suggests: kdebase Conflicts: koffice-libs (<< 1:1.1-20010404-0) Description: a chart drawing program for the KDE Office Suite KChart is a chart drawing application. . This package is part of the KDE Office Suite. > But in Packages.gz at the URL ftp://kde.debian.net/pub/kde/dists/potato/binary-i386 are new versions mentioned. ... Package: kchart Version: 2.0.1-0.potato2 Priority: optional Section: x11 Maintainer: Ivan E. Moore II <[EMAIL PROTECTED]> Depends: kdelibs3 (>= 4:2.0.1-0), koffice-libs (>= 2.0-final-0), libc6 (>= 2.1.2), libjpeg62, libpng2, libqt2.2 (>= 2:2.2.2-0) | libqt2.2-gl (>= 2:2.2.2-0), libstdc++2.10, libz1, xlib6g (>= 3.3.6-4), koffice-libs (>= 2.0.1-0.potato2) Suggests: kdebase Replaces: koffice-common, koffice-libs (<< 2.0.1-0.potato2) Architecture: i386 Filename: dists/potato/kde2/binary-i386/x11/kchart_2.0.1-0.potato2_i386.deb Size: 245990 MD5sum: 4367a5ea30764e993cb89ace8d7ccf87 Description: KDE Office Suite - KChart KDE Office Suite - KChart . A Graphical Chart program based on QT and KDE installed-size: 968 source: koffice ... What I'm missing or doing wrong? Matth
Re: koffice
> Version: 1:1.1-beta1-0.potato2 > Version: 2.0.1-0.potato2 You're fine, you have the latest version. 1.1-beta1 is actually newer than 2.0.1. Originally koffice was numbered the same as KDE (2.0.x), and then it went onto a separate release schedule with its own numbering system (1.0, 1.1). Ben.
