| 1 |
"Boyd Stephen Smith Jr." <bss03@××××××××××.net> posted |
| 2 |
200605201333.26654.bss03@××××××××××.net, excerpted below, on Sat, 20 May |
| 3 |
2006 13:33:21 -0500: |
| 4 |
|
| 5 |
> AFAIK, [confcache i]s still experimental. I'm currently blaming it for |
| 6 |
> some meta-kde emerge failures, but I also was using a number of other |
| 7 |
> experimental FEATURES at the same time, so I don't have any hard |
| 8 |
> evidence that it's broken. |
| 9 |
|
| 10 |
LOL. I seem to have some of the best luck with KDE... When everybody was |
| 11 |
saying it was broken with -Os, I was emerging it and having no issues. |
| 12 |
When everyone said it didn't work with the visibility stuff from gcc-4 |
| 13 |
backported to 3.4, I had no issues (IIRC in that case I didn't use the two |
| 14 |
apps known to have problems <shrug>). When it wasn't supposed to be |
| 15 |
working with gcc-4, guess what, I was running it compiled using gcc-4 |
| 16 |
(tho there were a couple niggles, but they were able to be worked around). |
| 17 |
|
| 18 |
Now, no real surprise, confcache all the way, here, and I routinely do up |
| 19 |
to five, sometimes more when it's KDE and it all came out at once, merges |
| 20 |
in parallel, using the still masked gcc-4 and now 4.1, now with confcache, |
| 21 |
and what would you know, no issues to speak of! (I'm having a slight |
| 22 |
issue with kwin's kompmgr, but that's due to the xorg-7.1-rcs I'm running, |
| 23 |
and again, it's easily controlled or I'd simply turn off composite. The |
| 24 |
parallel merges I use emerge --pretend --tree to get an overview of the |
| 25 |
dependency structure, then emerge --ask, verifying the same packages are |
| 26 |
not going to be merging in two separate konsole tabs at the same time.) |
| 27 |
|
| 28 |
>> > Of course there's ccache as well, but that can actually take longer |
| 29 |
>> > if you aren't doing much recompiling of the same code with the same |
| 30 |
>> > options on the same gcc, due to the extra disk activity of managing |
| 31 |
>> > the ccache, and there's distcc, if you have several computers that |
| 32 |
>> > can share the load. |
| 33 |
>> |
| 34 |
>> So it sould like this is some thing that is not needed and does not do |
| 35 |
>> much for the system in my case at less? |
| 36 |
> |
| 37 |
> Well, IIRC, ccache is able to match on file /data/ and basically ignores |
| 38 |
> the file /name/ which allows it some speed up some autoconf tests. Many |
| 39 |
> autoconf tests are just attempts to compile simple, but specially |
| 40 |
> constructed, C files. |
| 41 |
|
| 42 |
It does do that, I guess. |
| 43 |
|
| 44 |
> Also, ccache can speed up restarting packages that fail to emerge the |
| 45 |
> first time through. |
| 46 |
|
| 47 |
This was actually why I merged it here. Back when I started with Gentoo, |
| 48 |
I had bad memory -- rated PC3200 but good only to PC3000. Unfortunately, |
| 49 |
at the time, my board couldn't limit the memory clock in BIOS as the |
| 50 |
previous one I had had could do (the feature was later added in a BIOS |
| 51 |
update, and I eventually upgraded memory), so the system would work fine |
| 52 |
for days as long as it wasn't too stressed, but do a long compile and it |
| 53 |
would almost certainly crap out in the middle. ccache save my butt and |
| 54 |
my sanity, and made it possible, with patience, to run Gentoo and keep |
| 55 |
updated on a system that otherwise never could have done so. |
| 56 |
|
| 57 |
BTW, ccache and confcache are two different things, both of which can |
| 58 |
speed up merges, using different technologies. Just in case someone was |
| 59 |
confused and thinking they were the same things. |
| 60 |
|
| 61 |
|
| 62 |
|
| 63 |
-- |
| 64 |
Duncan - List replies preferred. No HTML msgs. |
| 65 |
"Every nonfree program has a lord, a master -- |
| 66 |
and if you use the program, he is your master." Richard Stallman |
| 67 |
|
| 68 |
-- |
| 69 |
gentoo-amd64@g.o mailing list |
Archives