Re: Depends and build depends in receipt

[ Thread Index | Date Index | More lists.tuxfamily.org/slitaz Archives ]


Hi Christopher,

> I have a question about the recompressing sources with tazdev.
> 
> How i'm i going to be able to just run one line command for
> downloading and recompressing the source? This was the reason we have
> tazwok get-src package. It will download the source if it doesn't
> exist and recompress it.

# cook pkgname --getsrc

Do you want something like --getsrc-all and a loop ?

> I still think its better for tank resources to do this when you get
> the newer sources instead of doing it in bulk. It could take a very
> long time doing it that way.

Yes if all src are not yet recompressed, on Tank they are and we just
have to recompress new source when needed.

For me the main raison to keep tar.bz2 is that on a standard slitaz
base or desktop you can extract archive via right-click and without
installing any additionnal packages, wich is not the case with lzma.

> Also i have a library that i'm working on so we can at least at the
> option to use tazwok depends check for cookutils.
> 
> https://bitbucket.org/godane/my-cookutils/changeset/2873f2b2384b
> 
> Changes the wok variable to point to your wok. Then put this in a tmp
> folder and type ./libcookorder gen-wok-db. This will create the
> wok-depends.txt and wok-wanted.txt and cookorder.txt. The
> cookorder.txt takes the longest to make.

real	3m 54.94s
user	2m 44.28s
sys	0m 49.44s
 
> Even if we can't use this code for tazwok cook-toolchain we can still
> use it to make it easier to make custom source dvd. This will allow
> us to know what depends are needed to rebuild everything and to make
> custom wok's for the custom source dvd too. NOTE: we will still have
> the full wok repos on disk. We will just copy out the folders needed
> for building everything.

I dont think these 312 lines will get into cookutils since the goal of
the tools is not to be able to rebuils all slitaz from scratch in one
command. On Tank and official packages we should have to rebuild
everthing only once a year when toolchain change and even at this time
I prefer doing it semi-manually to control one of the most important of
the distro.

If you realy miss a scratch function in cookutils, I can add a
wrapper... a command 'scratch' who will execute tazwok :-)

Also, using DEPENDS/BUILD_DEPENDS/WANTED to creat cookorder.txt is
realy complicated I think and I'm not sure about the consitency of the
list since it mess between deps and build deps (build deps should gave
priority. Look to the list produced on my system:

----
gawk
linux-api-headers
glibc
zlib
binutils
gmp
elfutils
mpfr
mpc-library
gcc
gdbm
db
perl
openssl
ncursesw
ncurses
bzip2
....

even the toolchain is not well ordered, gawk needs glibc-gcc to be
build as well as zlib and all above gcc. Here the toolchain should be
the first to be build if you want to follow LFS and be clean. It's
clear that you need base packages to start the first tmp toolchain,
like gcc, glibc and gawk is part of this base.

Also why *-dev packages are not build ? If zlib build and you want to
use it right after to build other packages you need to auto install it
as well as zlib-dev. Tazwok actually dont auto install freshly cooked
packages since I must update the chroot on Tank if I want it to use
latest packages.

> I really like cookutils but i want it to meet both of are needs with
> out use fighting each other or folking if its possible.

I dont want cookutils become tazwok, I also want to satisfy our bith
needs why I though leaving develop tazwok as you want and make
cookutils for Tank, me and contributor who wants. I also dont want to
fught ans it was never my intention, it's not cookutils vs tazwok,
there ar 2 different tools with different capacity but produce bothe
the same slitaz packages.

- Christophe

---
SliTaz GNU/Linux Mailing list - http://www.slitaz.org/


Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/