Is there a specific reason why dash and mawk isn't supported?

Zachary Kotlarek zach at
Thu May 5 13:49:53 PDT 2011

On May 5, 2011, at 3:19 PM, Erik Blomqvist wrote:

> Hi,
> I hope this is the right mailing list for this question. If nothing else you seem to be the right people to ask.
> Considering that dash and mawk are smaller and faster than bash and gawk I was a bit surprised to find that LFS doesn't support them. Even Ubuntu, that is a huge distribution by comparison, uses those packages because they provide better performance. Considering that one of the reasons for building your own linux system is to get better performance, it would make sense to use the best performing packages. So why doesn't LFS use or at least support these packages?
> I'm specifically interested in knowing if there are any technical reasons for not supporting these packages, e.g. package x doesn't work with dash/mawk. If it's just for historical reasons, maybe it's time to reconsider?

mawk is right out because at build time glibc assumes that awk is gawk and Drepper won't accept patches to allow any other behavior. (At least last time I checked -- admitted that was a couple of years ago). So distros can get away with replacing gawk but LFS can't, at least if it wants to be self-hosting.


You could make an argument for dash. It will cause problems with scripts/makefiles/etc. that assume /bin/sh is bash. I would agree that those scripts are incorrect and should just call bash directly, but I'm not sure it's worth the incompatibility.

IMHO, if you want to use dash you should start your scripts with /bin/dash and not /bin/sh. Then you can get the speed of dash when you want it without mucking about with the default shell. It's not like boot scripts (which I assume is the primary interest for dash) are highly portable and likely to be used unmodified on another distribution.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2746 bytes
Desc: not available
URL: <>

More information about the lfs-dev mailing list