Re: [gentoo-user] initial compile time
On 29/08/2017 21:30, Dale wrote: > Mick wrote: >> On Tuesday, 29 August 2017 12:08:12 BST Alan McKinnon wrote: >> PS. Did I mention you would have to do all of the above using a dial-up >> modem, >> which would require off-tree modules, because your PC came with a Winmodem? >> >> PPS. In case you have never been cursed with using a winmodem, check the 2nd >> bullet point at the bottom of this page: >> >> http://www.columbia.edu/kermit/winmodem.html >> > > > Could we not mention dial-u? It brings back bad memories. LOL I used > to hate when OOo would update, which took about 26 to 28 hours to > download, only to have it update again three or four days later. > Obviously, none of the previous tarball could be used. :-@ > > I've never done a stage 1 install but if I had a decent spare rig to > play with, I'd give it a shot. I'd just try not to shoot myself in the > foot. o_O No Dale, please actually go and do it. All of Gentoo will appreciate it, because you will be the guy that finds every last bug still hiding in the process. You know this will happen, so we are counting on you! -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
On 29/08/2017 18:47, Mick wrote: > On Tuesday, 29 August 2017 12:08:12 BST Alan McKinnon wrote: >> On 28/08/2017 22:20, mad.scientist.at.la...@tutanota.com wrote: >>> |it's not so hard yet, just hard to get by reading it in advance, doing >>> >>> it is fairly straight forward. >> >> It only seems like it might be tricky when reading it. >> Running Gentoo involves steps that you do on almost no other distro, so >> the words on the web page are unfamiliar. >> >> Almost everyone that's done it for the first time in the last 5 years >> and reported up says "Gee, that wasn't so hard! A lot easier than I ever >> thought it would be!" >> >> It's also very obvious what each step actually did, but you have to do >> it yourself once - it's obvious in petrospec. >> >> Welcome to Gentoo-land. May your stay be long and happy and fruitful. >> >> Your next assigned task (to be done sometime in the next year) is to >> repeat the whole process on a test rig and do it from a stage 1. Like we >> all did 15 years ago. If serious street cred brownie points is what you >> are after, that's how you get 'em! > > Nope. You need to do a stage 1 on nothing newer than a Pentium 3 or > equivalent for it to count, run out of compile space on PORTAGE_TMPDIR, > reformat and start again. > > For extra street cred you would of course install ~arch and have to work- > around every darn bug you came across by the n'th time you had to start > recompiling from scratch, because both gcc and the profile USE flags changed > 3 > times during the 2 weeks it would invariably take you to finish your install. > > In between all this pain, you may also discover your MoBo would hard lock > every time your mismatched RAM modules started paging over to swap ... O_O > > but after a hard reboot it would be OK until the next big compile. > > I understand this to be an endeavour not dissimilar to climbing Everest > during > an avalanche. > > PS. Did I mention you would have to do all of the above using a dial-up > modem, > which would require off-tree modules, because your PC came with a Winmodem? > > PPS. In case you have never been cursed with using a winmodem, check the 2nd > bullet point at the bottom of this page: > > http://www.columbia.edu/kermit/winmodem.html > Let's see, the most bottom line stage 1 I ever did was on an AMD athlon cpu with 512M Ram and an 80G IDE Drive. Didn't have the mobo lockups tho, but did have the devil's own time fiddling with overclocking and other goodies in the BIOS. Eventually I just selected "Optimal" and let them damn thing figure out what it thought was best... Sadly, I missed the whole winmodem saga and never had a need to use one. I had a US Robotics 56k dialup modem and in those days almost all mobos came with serial ports. So I probably got lucky. there's also the fun and games 20 years before that typing silly games on a ZX Spectrum, saving to cassette tape and hoping the bloody thing plays back OK. But that's a whole other story and I'm sure you are not even vaguely interested :-) -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
Mick wrote: > On Tuesday, 29 August 2017 12:08:12 BST Alan McKinnon wrote: >> On 28/08/2017 22:20, mad.scientist.at.la...@tutanota.com wrote: >>> |it's not so hard yet, just hard to get by reading it in advance, doing >>> >>> it is fairly straight forward. >> It only seems like it might be tricky when reading it. >> Running Gentoo involves steps that you do on almost no other distro, so >> the words on the web page are unfamiliar. >> >> Almost everyone that's done it for the first time in the last 5 years >> and reported up says "Gee, that wasn't so hard! A lot easier than I ever >> thought it would be!" >> >> It's also very obvious what each step actually did, but you have to do >> it yourself once - it's obvious in petrospec. >> >> Welcome to Gentoo-land. May your stay be long and happy and fruitful. >> >> Your next assigned task (to be done sometime in the next year) is to >> repeat the whole process on a test rig and do it from a stage 1. Like we >> all did 15 years ago. If serious street cred brownie points is what you >> are after, that's how you get 'em! > Nope. You need to do a stage 1 on nothing newer than a Pentium 3 or > equivalent for it to count, run out of compile space on PORTAGE_TMPDIR, > reformat and start again. > > For extra street cred you would of course install ~arch and have to work- > around every darn bug you came across by the n'th time you had to start > recompiling from scratch, because both gcc and the profile USE flags changed > 3 > times during the 2 weeks it would invariably take you to finish your install. > > In between all this pain, you may also discover your MoBo would hard lock > every time your mismatched RAM modules started paging over to swap ... O_O > > but after a hard reboot it would be OK until the next big compile. > > I understand this to be an endeavour not dissimilar to climbing Everest > during > an avalanche. > > PS. Did I mention you would have to do all of the above using a dial-up > modem, > which would require off-tree modules, because your PC came with a Winmodem? > > PPS. In case you have never been cursed with using a winmodem, check the 2nd > bullet point at the bottom of this page: > > http://www.columbia.edu/kermit/winmodem.html > Could we not mention dial-u? It brings back bad memories. LOL I used to hate when OOo would update, which took about 26 to 28 hours to download, only to have it update again three or four days later. Obviously, none of the previous tarball could be used. :-@ I've never done a stage 1 install but if I had a decent spare rig to play with, I'd give it a shot. I'd just try not to shoot myself in the foot. o_O Dale :-) :-)
Re: [gentoo-user] initial compile time
On Tuesday, 29 August 2017 12:08:12 BST Alan McKinnon wrote: > On 28/08/2017 22:20, mad.scientist.at.la...@tutanota.com wrote: > > |it's not so hard yet, just hard to get by reading it in advance, doing > > > > it is fairly straight forward. > > It only seems like it might be tricky when reading it. > Running Gentoo involves steps that you do on almost no other distro, so > the words on the web page are unfamiliar. > > Almost everyone that's done it for the first time in the last 5 years > and reported up says "Gee, that wasn't so hard! A lot easier than I ever > thought it would be!" > > It's also very obvious what each step actually did, but you have to do > it yourself once - it's obvious in petrospec. > > Welcome to Gentoo-land. May your stay be long and happy and fruitful. > > Your next assigned task (to be done sometime in the next year) is to > repeat the whole process on a test rig and do it from a stage 1. Like we > all did 15 years ago. If serious street cred brownie points is what you > are after, that's how you get 'em! Nope. You need to do a stage 1 on nothing newer than a Pentium 3 or equivalent for it to count, run out of compile space on PORTAGE_TMPDIR, reformat and start again. For extra street cred you would of course install ~arch and have to work- around every darn bug you came across by the n'th time you had to start recompiling from scratch, because both gcc and the profile USE flags changed 3 times during the 2 weeks it would invariably take you to finish your install. In between all this pain, you may also discover your MoBo would hard lock every time your mismatched RAM modules started paging over to swap ... O_O but after a hard reboot it would be OK until the next big compile. I understand this to be an endeavour not dissimilar to climbing Everest during an avalanche. PS. Did I mention you would have to do all of the above using a dial-up modem, which would require off-tree modules, because your PC came with a Winmodem? PPS. In case you have never been cursed with using a winmodem, check the 2nd bullet point at the bottom of this page: http://www.columbia.edu/kermit/winmodem.html -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] initial compile time
On Tue, Aug 29, 2017 at 1:50 AM, J García wrote: > > I would recomed using something like the base livecd or systemrescuecd > for an install with OpenRC. and only use something like CentOS(if you > are talking about 7) if you want to use systemd, and then make use of > systemd-nspawn and not chroot and bind mounts, and the systemd stage3. > As far as I can tell there are no problems if you install an openrc-based system using a systemd-based install CD, or a systemd-based system using an openrc-based install CD. You're not actually running init or services inside your chroot/nspawn-container/etc, so there really is no interaction with whatever init you're using outside. Now, if your install CD is running SELinux or something like that I could see there being issues, or if it is missing things like tar/etc. -- Rich
Re: [gentoo-user] initial compile time
On 28/08/2017 22:20, mad.scientist.at.la...@tutanota.com wrote: > |it's not so hard yet, just hard to get by reading it in advance, doing > it is fairly straight forward. > | It only seems like it might be tricky when reading it. Running Gentoo involves steps that you do on almost no other distro, so the words on the web page are unfamiliar. Almost everyone that's done it for the first time in the last 5 years and reported up says "Gee, that wasn't so hard! A lot easier than I ever thought it would be!" It's also very obvious what each step actually did, but you have to do it yourself once - it's obvious in petrospec. Welcome to Gentoo-land. May your stay be long and happy and fruitful. Your next assigned task (to be done sometime in the next year) is to repeat the whole process on a test rig and do it from a stage 1. Like we all did 15 years ago. If serious street cred brownie points is what you are after, that's how you get 'em! -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
2017-08-28 14:20 GMT-06:00 : > > used about 3 lines of USE flags, actually it was less than a half hour, which > doesn't surprise me if the stage 3 is compiled. i am using a fairly quick > drive ;(and i was watching vid) had 2 small problems but i'll try again off > the live dvd. You mean 3 lines of ~80 characters of USE flags in make.conf? That is a lot and shurely will come and bite you in the future. I guess most of us learn a sane way to handle use flags the hard way, just sharing, these days my USE flags in make.conf are less than one line even less than 8 flags, I would say a good aproach is to select the right profile for your need is the most important, check the flags that it has(emerge --info), and then only the flags that you really really want on *every* package you put in make.conf, everything else goes to /etc/portage/package.use/my.use on a per package basis, and everything else portage finds out it needs to build what i want goes to /etc/portage/package.use/zzz_auto.use. And keep your world to only what you need, learn about --oneshot as soon as you can. i got an error with the rbind "mount --rbind /sys /mnt/gentoo/sys" i should have guessed the stage3 was just compressed, > and it also couldn't find a "sane" /dev. probably both due to centos not > playing nice (and my mounting and unmounting of a couple partitions). I would recomed using something like the base livecd or systemrescuecd for an install with OpenRC. and only use something like CentOS(if you are talking about 7) if you want to use systemd, and then make use of systemd-nspawn and not chroot and bind mounts, and the systemd stage3.
Re: [gentoo-user] initial compile time
On Mon, Aug 28 2017, Rich Freeman wrote: > On Mon, Aug 28, 2017 at 2:37 PM, allan gottlieb wrote: >> On Mon, Aug 28 2017, Alan McKinnon wrote: >> >>> I forgot to mention there are things that /significantly/ improve >>> compile times. Top of the list is /var/tmp/portage on tmpfs. >> >> Does this mean that, if the build fails, you build again not on tmpfs so >> as to capture build.log? >> > > Just copy the build log off the tmpfs. Portage doesn't remove it > automatically. Unless you reboot the system before inspecting it the > file will still be there unless you have a really overzealous > tmpreaper. > > (Or if you're doing the build in a systemd unit. I actually build > binary packages overnight from a unit and was always puzzled about why > the build was gone from /var/tmp/portage when I'd inspect errors. > Then I realize the default is for units to have a private /var/tmp - > so it was gone when the unit terminated. That can of course be > disabled.) Thank you Rich and Joost. allan
Re: [gentoo-user] initial compile time
excellent idea, thanks. -- The Power Of the People Is Stronger Than The People In Charge. 28. Aug 2017 12:45 by jo...@antarean.org: > On 28 August 2017 20:37:30 GMT+02:00, allan gottlieb <> gottl...@nyu.edu> > > wrote: > >On Mon, Aug 28 2017, Alan McKinnon wrote: >>> I forgot to mention there are things that /significantly/ improve >>> compile times. Top of the list is /var/tmp/portage on tmpfs. > >Does this mean that, if the build fails, you build again not on tmpfs > >so > >as to capture build.log? > >thanks, > >allan > > As long as you don't reboot, the entire contents will remain. > > If you set PORT_LOGDIR, the build.log will be copied to the specified > location even if the build fails. > > -- > Joost > -- > Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] initial compile time
good tip, thanks. -- The Power Of the People Is Stronger Than The People In Charge. 28. Aug 2017 12:55 by ri...@gentoo.org: > On Mon, Aug 28, 2017 at 2:37 PM, allan gottlieb <> gottl...@nyu.edu> > wrote: >> On Mon, Aug 28 2017, Alan McKinnon wrote: >> >>> I forgot to mention there are things that /significantly/ improve >>> compile times. Top of the list is /var/tmp/portage on tmpfs. >> >> Does this mean that, if the build fails, you build again not on tmpfs so >> as to capture build.log? >> > > Just copy the build log off the tmpfs. Portage doesn't remove it > automatically. Unless you reboot the system before inspecting it the > file will still be there unless you have a really overzealous > tmpreaper. > > (Or if you're doing the build in a systemd unit. I actually build > binary packages overnight from a unit and was always puzzled about why > the build was gone from /var/tmp/portage when I'd inspect errors. > Then I realize the default is for units to have a private /var/tmp - > so it was gone when the unit terminated. That can of course be > disabled.) > > > > -- > Rich
Re: [gentoo-user] initial compile time
oh ya, i' definetly watch like a hawk evntually. i'm using all cat 7 wiring, with different network cards i could go 10G between my machines. I went with cat 7, mostly to make sure it was screened properly, and some what future proof. -- The Power Of the People Is Stronger Than The People In Charge. 28. Aug 2017 14:09 by alan.mckin...@gmail.com: > On 28/08/2017 18:55, John Blinka wrote: > > >> invariably works. I'm not criticizing your speedup plans - by all >> means, have fun - but if you're just starting out in Gentoo, be aware >> that these speedups aren't necessarily a slam-dunk. > > > OP should also watch out for the Gentoo ricer curse syndrome > > > -- > Alan McKinnon > alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
used about 3 lines of USE flags, actually it was less than a half hour, which doesn't surprise me if the stage 3 is compiled. i am using a fairly quick drive ;(and i was watching vid) had 2 small problems but i'll try again off the live dvd. i got an error with the rbind "mount --rbind /sys /mnt/gentoo/sys" i should have guessed the stage3 was just compressed, and it also couldn't find a "sane" /dev. probably both due to centos not playing nice (and my mounting and unmounting of a couple partitions). it's nice to know i can make some progress before i have my compiler farm up, i'm sure i'll want it before it's up (one machine to finish upgrading, one with bad power supply caps for the graphics system (easy, just hard to get it apart enogh to find out when it's in a tiny case. i'll doubtless be doing some bench marks, especially if i upgrade the cpu. my main machine has 4G of ram, i've rarely used more than 1G, usually less but this should give me a chance to exercise it well. it's not so hard yet, just hard to get by reading it in advance, doing it is fairly straight forward. it's a stage3 that's about 2-3 weeks old now, but i did let it sync and download the newest stuff, and it did a fair amount of downloading (it's just amazing what gigabit fiber to premises can do, especially when the city runs it. people who got it right away after the cable was in get it for $50/month, for life even if they move as long as they stay in longmont. and no deposit ,or install fees. the 'ONT' can be up to $300 dollars. i keep mine on top of a bookcase! i love it! hope we all have it soon. i'm even allowed to run a server, though that may be a good winter project. if i run four machines in a farm, and amuse my self with the mac i'll obviously want to do it when it's cool. no guilt in winter, it's electric baseboard, as expensive as possible but it's rolled into the rent.(and it's a very drafty apartment) point being that nearly all the electric baseboards will just run less often with 5 machines up, till then it's probably a night project, my best hours. after all you can be certain most of the electricity someone uses is very nearly converted, 100% to heat, doesn't hurt to do something with it with it on it's conversion to heat. i'm still using a 100Mbps router, also trying to learn how to use pfsense, fortunately i've studied that a bit more, as far as how the new works, and i've agresively operated a software firewall before (1999, with a barely working phone line, had a great 56K modem, with a display to tell you when it changes the up/down speed. it oscillated between about 20 and ocasionaly 26K. that's what happens when 7 lines go to the development, and get demultiplexed into 100 phone lines. it was pretty good for audio, but 56K modems use an incredibly complex line testing and correction. -- The Power Of the People Is Stronger Than The People In Charge. 28. Aug 2017 10:08 by alan.mckin...@gmail.com: > On 28/08/2017 13:41, > mad.scientist.at.la...@tutanota.com> wrote: >> Ok, i'm starting to understand the install instructions, a steeper curve >> than i expected but still way easier than LFS. >> So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long >> will stage3 take to compile, roughly? > > You don't compile a stage 3, that has already been done and you > downloaded it. You bunzip2 and untar a stage 3. > > F > > -- > Alan McKinnon > alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
On 28/08/2017 18:55, John Blinka wrote: > invariably works. I'm not criticizing your speedup plans - by all > means, have fun - but if you're just starting out in Gentoo, be aware > that these speedups aren't necessarily a slam-dunk. OP should also watch out for the Gentoo ricer curse syndrome -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
On Mon, Aug 28, 2017 at 2:37 PM, allan gottlieb wrote: > On Mon, Aug 28 2017, Alan McKinnon wrote: > >> I forgot to mention there are things that /significantly/ improve >> compile times. Top of the list is /var/tmp/portage on tmpfs. > > Does this mean that, if the build fails, you build again not on tmpfs so > as to capture build.log? > Just copy the build log off the tmpfs. Portage doesn't remove it automatically. Unless you reboot the system before inspecting it the file will still be there unless you have a really overzealous tmpreaper. (Or if you're doing the build in a systemd unit. I actually build binary packages overnight from a unit and was always puzzled about why the build was gone from /var/tmp/portage when I'd inspect errors. Then I realize the default is for units to have a private /var/tmp - so it was gone when the unit terminated. That can of course be disabled.) -- Rich
Re: [gentoo-user] initial compile time
On 28 August 2017 20:37:30 GMT+02:00, allan gottlieb wrote: >On Mon, Aug 28 2017, Alan McKinnon wrote: > >> I forgot to mention there are things that /significantly/ improve >> compile times. Top of the list is /var/tmp/portage on tmpfs. > >Does this mean that, if the build fails, you build again not on tmpfs >so >as to capture build.log? > >thanks, >allan As long as you don't reboot, the entire contents will remain. If you set PORT_LOGDIR, the build.log will be copied to the specified location even if the build fails. -- Joost -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] initial compile time
On Mon, Aug 28 2017, Alan McKinnon wrote: > I forgot to mention there are things that /significantly/ improve > compile times. Top of the list is /var/tmp/portage on tmpfs. Does this mean that, if the build fails, you build again not on tmpfs so as to capture build.log? thanks, allan
Re: [gentoo-user] initial compile time
On Mon, Aug 28, 2017 at 7:41 AM, wrote: > Ok, i'm starting to understand the install instructions, a steeper curve > than i expected but still way easier than LFS. > So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long will > stage3 take to compile, roughly? With emphasis on the "roughly", I seem to recall that it took me several days to compile everything I wanted on initial system installation on machines of similar capabilities. But that includes significantly more than stage3, with long build time stuff like firefox, chromium, and libreoffice accounting for at least half that time. > > next month i'll be setting up a compiler farm with 3 other, similiar > machines which should help, will also be upgrading cpus to 4 or 6 core, and > have one machine that can upgraded for phenom and one i can update to > opteron, according to the board makers (it just needs a different bios). > > If i had the supporting bios any of my machines could upgrade to nearly any > AMD socket 2 or 3+ chip You seem to want to see how fast you can go, and that's certainly an interesting exercise. I've been there, too, but over the years have gradually retreated to a very non-aggressive compile setup. I've used distcc, made use of multiple cores, parallel make, etc. but have abandoned them all over time. It is quite possible to slow things down with improper setup, or with a local network with limited capabilities, so it takes a little time and experimentation to tune things properly. And it's possible to speed things up substantially, too. However, in my experience, speedups obtained this way can and do expose bugs in the build process. For me the personal keyboard time I invested in fixing things that broke in parallel wasn't worth the speedups I achieved. I, too, come from the punchcard and paper tape era, so even the very cheapest modern cpus run circles around the multi-million $ parallel supercomputers I used to buy and use. I now prefer just starting an emerge, and letting it take its merry old time. Gentoo's gotten good enough over the years that this almost invariably works. I'm not criticizing your speedup plans - by all means, have fun - but if you're just starting out in Gentoo, be aware that these speedups aren't necessarily a slam-dunk. John Blinka
Re: [gentoo-user] initial compile time
On 28/08/2017 18:47, Rich Freeman wrote: > On Mon, Aug 28, 2017 at 9:08 AM, Alan McKinnon > wrote: >> On 28/08/2017 13:41, mad.scientist.at.la...@tutanota.com wrote: >>> Ok, i'm starting to understand the install instructions, a steeper curve >>> than i expected but still way easier than LFS. >>> So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long >>> will stage3 take to compile, roughly? >> >> How many extra USE flags did you set? What software did you decide to >> add? How old is the stage 3 you used? A recent one will obviously need >> fewer recompiles than a 6-month old one. But to portage it's all the >> same - make a list of stuff to be done and do it from the beginning. >> > > I imagine that some people like to rebuild everything after changing > their CFLAGS, but there is certainly no requirement to do so. > > Also, I tend to favor getting the system booting on its own before > fiddling with it. There isn't much you need to build before you have > the system booting, and at that point you're at least running while > you're doing the rest. If I'm going to switch the init implementation > then I'll do that before the first boot since that impacts the grub > configuration. Systemd does take a bit to build, but it is still a > far cry from doing a ton of rebuilds. > > However, if you do have a /etc/portage configuration you like, and a > /var/lib/portage/world, then you can just update these and the profile > and just do an emerge -NDu world and watch it all come into place. I > have a typical set of configuration files I deploy on any new system > so I will often build this stuff directly before rebooting (which > includes stuff like systemd, vim, screen, grub, dracut, and so on). > Good points Rich. I forgot to mention there are things that /significantly/ improve compile times. Top of the list is /var/tmp/portage on tmpfs. On 16G Ram and 8G tmpfs, I see total build times usually halved. firefox, thunderbird, libreoffice, icu and webkit all benefit, and those packages seem to be updating more often than most these days as well as being the biggest time hogs -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] initial compile time
On Mon, Aug 28, 2017 at 9:08 AM, Alan McKinnon wrote: > On 28/08/2017 13:41, mad.scientist.at.la...@tutanota.com wrote: >> Ok, i'm starting to understand the install instructions, a steeper curve >> than i expected but still way easier than LFS. >> So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long >> will stage3 take to compile, roughly? > > How many extra USE flags did you set? What software did you decide to > add? How old is the stage 3 you used? A recent one will obviously need > fewer recompiles than a 6-month old one. But to portage it's all the > same - make a list of stuff to be done and do it from the beginning. > I imagine that some people like to rebuild everything after changing their CFLAGS, but there is certainly no requirement to do so. Also, I tend to favor getting the system booting on its own before fiddling with it. There isn't much you need to build before you have the system booting, and at that point you're at least running while you're doing the rest. If I'm going to switch the init implementation then I'll do that before the first boot since that impacts the grub configuration. Systemd does take a bit to build, but it is still a far cry from doing a ton of rebuilds. However, if you do have a /etc/portage configuration you like, and a /var/lib/portage/world, then you can just update these and the profile and just do an emerge -NDu world and watch it all come into place. I have a typical set of configuration files I deploy on any new system so I will often build this stuff directly before rebooting (which includes stuff like systemd, vim, screen, grub, dracut, and so on). -- Rich
Re: [gentoo-user] initial compile time
On 28/08/2017 13:41, mad.scientist.at.la...@tutanota.com wrote: > Ok, i'm starting to understand the install instructions, a steeper curve > than i expected but still way easier than LFS. > So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long > will stage3 take to compile, roughly? You don't compile a stage 3, that has already been done and you downloaded it. You bunzip2 and untar a stage 3. >From that point on, you will do exactly the same that you will do for every other world update on Gentoo you will ever do: - set USE flags as needed - set items in make.conf as needed - decide what software you want - tell portage to make it so The "make it so" part is your real question, but no-one can answer it as it's a "how long is a piece of string" question. How many extra USE flags did you set? What software did you decide to add? How old is the stage 3 you used? A recent one will obviously need fewer recompiles than a 6-month old one. But to portage it's all the same - make a list of stuff to be done and do it from the beginning. There is really no functional point in updating the stage 3 and then doing the above; unless you want to see how it all works. Might be nice to do it the first time, but you will likely never do it again :-) With that hardware, I'd thumb suck guess about 3 hours if the machine will be headless, and closer to 12 if it's a workstation with most of say KDE or Gnome, plus a browser and openoffice and an MUA etc etc etc. But that's just a guess, could also be anywhere from 1 hour on the low end to 24 on the high end. Or more. -- Alan McKinnon alan.mckin...@gmail.com
[gentoo-user] initial compile time
Ok, i'm starting to understand the install instructions, a steeper curve than i expected but still way easier than LFS. So, on a dual core athlon II 6000 (two cores, 3ghz) roughly how long will stage3 take to compile, roughly? next month i'll be setting up a compiler farm with 3 other, similiar machines which should help, will also be upgrading cpus to 4 or 6 core, and have one machine that can upgraded for phenom and one i can update to opteron, according to the board makers (it just needs a different bios). If i had the supporting bios any of my machines could upgrade to nearly any AMD socket 2 or 3+ chip, and i'll be looking for that and, i'll also be looking at the linux bios project to see if any of my hardware is well supported, may also find custom bioses that gamers have made. in any case, i'll be learning steep, which i love once i'm into it, and i'm about there with gentoo. i love the useflags idea, and i was looking for a good os for learning linux, considered slack and lfs, But gentoo looks good, and i like to get my hands dirty. couple of minor bugs in the handbook but not bad, also many links on other sites for gentoo.org with bad links, oh well, there always are. I do realize the labor involved, hope to give some back to the community. yes, i'm a computer nut, hardware and software, and the coming used machines all get a bath and the cases go into the shower with me, only big wet space in the place and haveing the head on a hose really helps. I've run a mac board in a modified pc case, off of 2 lab type power supplies, actually for ahile it was naked on the table, works great, solves cooling issues (not that a 68040 gets that hot). i've hand compiled assembly on a 6502 and a z80, the second one on a sinclair timex i put an external keyboard on and seperated the ram expansion board from the conector and velcroed it on top with some ribbon cable. i've actually used fortan a little on punch cards, on an old ibm 360 it think it was. had a card punch i got at auction for a bit. the old ibm punches are nearly all mechanical, just like a selectric typwriter. than again, i know a little cobol i can't make go away. -- The Power Of the People Is Stronger Than The People In Charge.
