Archaic wrote:
Randy has given some answers for BLFS, but let me reaffirm...
> 1a) What are we rounding to on single file sizes (like size of tarballs
> and patches)?
We use 'ls -lh'. 'du -k'/1024 also give the same value. Rounding for
both file size and build size is:
If < 1M, xxx KB
elseif < 10M x.y MB
elseif < 100M xx.y MB
eles xxx MB
> 1b) How should we determine a single file's size? The lfs edguide gives
> an example of ls -l and assumes the output will be in bytes, but
> aliases and/or env vars could screw with that. Also, the lfs edguide
> says to divide the number by 1024, ls -lk will do that for us if we
> are indeed rounding to kB, otherwise ls -l --block-size=1 divided by
> wherever we are rounding to will work.
As above.
> 2a) What are we rounding to on installed size?
As above. Build size includes space for both the build and the install.
> 2b) How are we getting installed size? Bruce, I noticed you do
> df -k $LFS | grep $LFS | sed -e "s/ \{2,\}/ /g" | cut -d' ' -f3, but
> df -k /x |grep /x | awk ' { print $3 }' is less expensive. Either
> way, df -k seems to be the most consistant and easiest to maintain.
The edguide is a guide. Alternative methods are allowed if equivalent.
Note that df works on partitions so df -k / and df -k /x are the same if
the build partiton is on the root partition. Other alternatives like
'du -sh' are OK, but both the build and the install locations need to be
taken into account.
> 3a) Where are we rounding SBU's to?
If < 0.1 use "less than 0.1 SBU"
else if less than 20SBU, xx.x SBU
else xx SBU.
Note that the book says: "Estimated build time"
> 3b) Does the *official* SBU account for tarball extraction all the way
> to removing the sources (with exception for the occasion where the
> sources aren't removed). If so, does LFS's expect page account for
> the removal of the tcl source dir since that is where it happens?
The method is documented at:
http://www.linuxfromscratch.org/~bdubbs/about.html
It does not include tarball extraction or directory cleanup. Note that
cleanup is really minimal compared to build time. Tarball extraction
does soemtimes take significant clock time, but it too is generally
pretty small compared to build time.
> 3c) Does the *official* SBU account for testsuites? Or should those be
> timed and listed separately in the explanatory text that mentions
> the testsuite? What about if the book doesn't "recommend" running a
> testsuite (which I think is generally insane, BTW)?
The build time does not include the teting time. Testing time is
mentioned in parenthetical remarks if appropriate for the package.
-- Bruce
--
http://linuxfromscratch.org/mailman/listinfo/blfs-dev
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
