| View previous topic :: View next topic |
| Author |
Message |
96140
Retired Dev
Joined: 23 Jan 2005
Posts: 1324
|
 Posted: Sun Mar 27, 2005 8:46 am Post subject: |
 |
|
| Sith_Happens wrote: | This is good advice, last time I installed I compiled with glibc-2.3.4-20050125, and then after upgrading to gcc 3.4.3, upgraded glibc to 2.3.4-20050125-r1. However, if 2.3.4-20050125 compiles, then I would suggest you stick with that version. Aside from some bug fixes, theres not a real reason to upgrade to r1. It's marked testing for a reason.  |
Yes, I just don't see the need to upgrade to 20050125-r1, as I haven't had any problems or bugs with the previous version. In fact, I don't believe I've had to upgrade a single toolchain-related package; I'm still using the original toochain I emerged and rebuilt using the guide: gcc 3.4.3.20050110, glibc 20050125, and a couple of others that I'll have to check once I'm on my Gentoo box.
The only major upgrade I made has been moving from kernel 2.6.10-gentoo-r6 to the latest gentoo-dev-sources kernel (2.6.11-r4). I try to stay away from apparently minor toolchain updates such as this, or until I hear that I'll gain a noticeable performance and/or reliability boost. Otherwise, I just have a lot of intensive unnecessary compiling to be doing at all times. Switching from gcc 3.3.x to 3.4.x is sufficient reason to upgrade and re-emerge the toolchain. Anything else probably isn't worth my time, and as my versions of gcc and related toolchain packages are completely stable, why would I need to upgrade? 
Thank goodness I noticed the -r1 update to glibc on my most recent emerge -auD world, otherwise I'd be making Bugzilla reports! Stick with the more reliable, (barely) older toolchain components in ~x86 for the time being, until this whole ebuild mess gets straightened out... |
|
| Back to top |
|
 |
Bob P
Advocate
Joined: 20 Oct 2004
Posts: 3355
Location: Jackass! Development Labs
|
| Posted: Sun Mar 27, 2005 9:46 am Post subject: |
|
|
| nightmorph wrote: | I try to stay away from apparently minor toolchain updates such as this, or until I hear that I'll gain a noticeable performance and/or reliability boost. Otherwise, I just have a lot of intensive unnecessary compiling to be doing at all times. Switching from gcc 3.3.x to 3.4.x is sufficient reason to upgrade and re-emerge the toolchain. Anything else probably isn't worth my time, and as my versions of gcc and related toolchain packages are completely stable, why would I need to upgrade?
|
well, if you can't honestly say that you have at least one, if not two, of your machines performing a major upgrade at any given point in time, then you can't honestly call yourself a basement-dwelling linux geek. turn in your beanie-copter and hang your head in shame! 
| nightmorph wrote: | | Thank goodness I noticed the -r1 update to glibc on my most recent emerge -auD world, otherwise I'd be making Bugzilla reports! Stick with the more reliable, (barely) older toolchain components in ~x86 for the time being, until this whole ebuild mess gets straightened out... |
based on your comments i don't think you understand what has happened. the problem that came along this week was not limited to 20050125-r1. if you read the bug reports, 20041102-r1 got whacked at the same time because Azarah and Vapier were revising the glibc ebuilds. unfortunately, changes to the ebuilds are not always reflected in their file names -- you have to look at their revision numbers.
i had three boxes compiling toolchain updates this week. each of them emerged glibc 2.3.4.200501225-r1, and only one of them got caught by the bug:
the first box (my main desktop) downloaded the glibc 2.3.4.20050125-r1 ebuild before the nefarious edit, and completed the emerge without a hitch. i've subsequently rebuilt the world on that box and everything is running fine. 
the second box started emerging the next day, caught the b0rked 20050125-r1 ebuild during its short lifespan, and encountered a critical error that stopped the toolkit rebuild two days into compling glibc.  (why do critical errors like this always seem to hit the slowest box, two days into the build?)
the third box started the toolkit rebuild yesterday, after the revised ebuilds hit the portage tree. it also compiled glibc 2.3.4.20050125-r1 and rebuilt the world files without any problems.
blanket references to 20050125-r1 being problematic simply are not accurate. 20050125-r1 was only bad for a couple of days. in addition, 20041102-r1 also went bad at the same time. based on its name you would think that 20041102-r1 is four months old! it also got borked this week, because both ebuilds were transiently subject to a buggy update. you woudn't expect a "4 month old" ebuild to transiently go bad for a couple of days, but there you go. the good news is that both ebuilds have been fixed.
if you want to get the full story, take a look at the ebuild revision histories in the CVS tree:
http://www.gentoo.org/cgi-bin/viewcvs.cgi/sys-libs/glibc/glibc-2.3.4.20041102-r1.ebuild?view=graph
http://www.gentoo.org/cgi-bin/viewcvs.cgi/sys-libs/glibc/glibc-2.3.4.20050125-r1.ebuild?view=graph
for everyone who's wondering: Yes, its safe to go into the water ... for now.
_________________
.
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
| Back to top |
|
|
Sith_Happens
Veteran
Joined: 15 Dec 2004
Posts: 1807
Location: The University of Maryland at College Park
|
| Posted: Sun Mar 27, 2005 2:52 pm Post subject: |
|
|
Thanks Bob. I upgraded to glibc-2.4.3.20050125-r1 the day it was marked "testing", so I'm not suprised that I wasn't hit by this buggy ebuild. However, I think in general n00b's should follow nightmorph's advice. Although the stage 1/3 install does require you to work a little in the testing branch of portage, that doesn't mean that they need to continually upgrade to the latest version of glibc everytime one is added to the testing branch. Instead it's really a better idea that they freeze the version of glibc they use with the ~x86 keyword by adding <=sys-libs/glibc-2.3.4.20050125-r1 ~x86 to /etc/portage/package.keywords, and wait until later versions become stable and/or their is an intelligent reason to upgrade. Doing the same with gcc 3.4.3-r1 would be a smart choice as well. Messing up glibc or gcc is pretty serious if you don't know enough to fix it, and that's just not a situation n00bs really need or want to get themselves in.
_________________
"That question was less stupid; though you asked it in a profoundly stupid way."
I'm the brains behind Jackass! | Tutorials: Shorewall |
|
| Back to top |
|
|
Bob P
Advocate
Joined: 20 Oct 2004
Posts: 3355
Location: Jackass! Development Labs
|
| Posted: Mon Mar 28, 2005 2:19 am Post subject: |
|
|
i'm not going to lose too much sleep over this. if a n00b decides to use an advanced method and gets himself in trouble, and can't find his way out of a paper bag with a blowtorch, then who is to blame? if they can't hande it, they can stick with a Stage 3 tarball and the default gcc 3.3.4.
as they say, "if you can't run with the big dogs, you might as well stay on the porch."
by the way, i got a personal message today from a n00b who just couldn't resist bothering me even though i've made my stance on personal support messages very clear.
he demaneded that since 2005.0 has been released, i should immediately upgrade the guide to be 2005.0 compliant. in addition, i was told to use a 686 host in the guide to make it easy for him to follow. in response, i'd like to publicly tell that jackass to go solve his own problems.
_________________
.
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
| Back to top |
|
|
Sith_Happens
Veteran
Joined: 15 Dec 2004
Posts: 1807
Location: The University of Maryland at College Park
|
| Posted: Mon Mar 28, 2005 2:59 am Post subject: |
|
|
| Bob P wrote: | | in response, i'd like to publicly tell that jackass to go solve his own problems. |
Here here. 
_________________
"That question was less stupid; though you asked it in a profoundly stupid way."
I'm the brains behind Jackass! | Tutorials: Shorewall |
|
| Back to top |
|
|
Bob P
Advocate
Joined: 20 Oct 2004
Posts: 3355
Location: Jackass! Development Labs
|
| Posted: Mon Mar 28, 2005 3:03 am Post subject: |
|
|
ok, i feel better now that i've ranted. 
i can honestly understand why the guys who want 2005.0 instructions are anxious, but i think its over the line of common courtesy for somebody to tell me what architecture i should use in the guide.
IMHO, a separate update to the guide isn't even necessary. just deal with the fact that you already have linux26-headers and udev in the tarball.
but for those people who can't live without a set of instructions to follow, i've posted a quick-and-dirty update here:
Stage 1/3 Guide for 2005.0 Installation Media.
This guide is experimental and is an open-source project. I'll let the community work the bugs out of this one.
_________________
.
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
| Back to top |
|
|
Bob P
Advocate
Joined: 20 Oct 2004
Posts: 3355
Location: Jackass! Development Labs
|
| Posted: Mon Mar 28, 2005 3:34 am Post subject: |
|
|
 on the subject of "Why Pentium?"
the guide uses pentium architecture on purpose -- its the common denominator for just about any modern x86-type PC that would be the subject for a Gentoo installation. regardless of which modern x86 descendant PC you may have, you could use "pentium" as it is written in the guide and the installation would run on just about any modern PC. that would not be the case if i used a different/more recent architecture as an example.
_________________
.
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
| Back to top |
|
|
witchdoctor
n00b
Joined: 18 Nov 2003
Posts: 3
|
| Posted: Mon Mar 28, 2005 4:27 am Post subject: |
|
|
| Awesome guide Bob. Should the hosts file read 127.0.0.1 instead of 127.0.1? Does the hostname -f work where you specified it? I had to do a reboot and use the old hostnaming method to get things going. Thanks for the guide. Doc |
|
| Back to top |
|
|
Sith_Happens
Veteran
Joined: 15 Dec 2004
Posts: 1807
Location: The University of Maryland at College Park
|
| Posted: Mon Mar 28, 2005 4:54 am Post subject: |
|
|
| witchdoctor wrote: | | 127.0.0.1 instead of 127.0.1 |
Yeah, that's a typo. | witchdoctor wrote: | | Does the hostname -f work where you specified it? |
If you update /etc/hosts, then it should show your hostname when you run hostname -f. It displays the first name in the list however, so if your localhost (127.0.0.1) line in /etc/hosts looks like this: | Code: | 127.0.0.1 localhost BlueBox
|
Then hostname -f will return localhost, and hostname will return BlueBox. If it looks like this: | Code: | | 127.0.0.1 BlueBox localhost |
Then both hostname and hostname -f will return BlueBox. In either case however, both are valid identifiers for your machine. Perhaps that is what is confusing you?
_________________
"That question was less stupid; though you asked it in a profoundly stupid way."
I'm the brains behind Jackass! | Tutorials: Shorewall |
|
| Back to top |
|
|
Bob P
Advocate
Joined: 20 Oct 2004
Posts: 3355
Location: Jackass! Development Labs
|
| Posted: Mon Mar 28, 2005 6:13 am Post subject: |
|
|
cripes! how long has that typo been there without anybody noticing it?!?  i'd think that by including "hostname -f" as an error check, somebody would have noticed the error! thanks for pointing that out.
as a matter of semantics, its probably not right to ever edit the localhost line in /etc/hosts to include your hostname alias for the individual PC. your actual hostname should always go on the line that specifies its unique IP address. i've seen some pretty heated discussions on this topic... 
_________________
.
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
| Back to top |
|
|
96140
Retired Dev
Joined: 23 Jan 2005
Posts: 1324
|
| Posted: Mon Mar 28, 2005 8:03 am Post subject: |
|
|
| Bob P wrote: | but for those people who can't live without a set of instructions to follow, i've posted a quick-and-dirty update here:
Stage 1/3 Guide for 2005.0 Installation Media.
This guide is experimental and is an open-source project. I'll let the community work the bugs out of this one. |
Jeez, Bob, aren't you going to burn yourself out further by posting this, even if you completely don't intend to put that much into it yourself? I think it's a great idea to just put out the basics and let others work with it from here on out; I just wonder if they (the newbies) will set the PM barrage to rapid-fire! We, the community, would hate to have you ultimately turned off to the rest of us, as you've put in a lot of time, trouble, and service--simply because some people don't get it, when it comes to issues like decency and the fact that you're not at our beck and call and not being paid for creativity!
*hands in beanie copter*
And I've learned something new, today! While some of my comments were misunderstood, I'll just move on, because I did learn. And no, I'm not a basement-dwelling Linux geek...it's too sunny and warm here in Southern California! But you still haven't sold me on having a major update compiling in the background. If the devs aren't putting out serious package changes, like huge bugfixes or feature updates, then there's not much to compile...is there? Sometimes the bleeding edge doesn't have enough major updates to make to keep everyone happy, I guess!
Thanks for the new (untested) 2005.0 Guide, though it just seems like a different Stage 3 tarball is grabbed. IIRC, the entire point of this 2004.3 Stage 1/Stage 3 guide was making use of a better toolchain, and it's a shame that the new (delayed) 2005 CD doesn't include the better toolchain packages, even though they're 'unstable'. I'd sure like to try a stage 1 install without having to use a Stage 3 tarball; wouldn't it be nice if bootstrap.sh ran perfectly on an 'unstable' toolchain? Mmm, gcc 3.4.3.x.....needs to be on an optional ~x86 CD!
But I digress. Back to you, Bob. |
|
| Back to top |
|
|
Deathwing00
Bodhisattva
Joined: 13 Jun 2003
Posts: 4087
Location: Dresden, Germany
|
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Installing Gentoo
