jhalfs experimental and ICA

Dan Nicholson dbn.lists at gmail.com
Thu Mar 9 11:54:29 PST 2006

On 3/9/06, M.Canales.es <manuel at linuxfromscratch.org> wrote:
> El Jueves, 9 de Marzo de 2006 20:07, Dan Nicholson escribió:
> > To address this situation, you could create a timestamp at the end of
> > the iteration.
> For libraries and binaries that could work. For files that aren't created in
> the make phase but copied from the source tree, maybe not.

Hmm, this is a tricky situation.  Fortunately, the libraries and
binaries are what we really care about, but you have a point.

> A time-based find on a fresh build could show if there is actualy some file of
> that type. If some is found, maybe a touch on that files could to solve the
> issue.

But how would we differentiate between a file that hasn't been updated
and a file with an old timestamp?  Here's one thought:

After iter1, create timestamp1.1.  Touch all the files.  We know
they're new or they wouldn't exist.  Create timestamp1.2.

After iter2, create timestamp2.1.  Perform following conditionals:  If
file is older than timestamp1.1 or newer than timestamp1.2, it has
been updated.  Touch it.  Else, file is newer than timestamp1.1, but
older than 1.2, and it is the same file.  Warn about file.  After all
files, create timestamp2.2.

Iterate, using the timestamps from the previous iteration.

This might be overkill, but I couldn't think of anything else.  In my
situation, I use DESTDIR's, so I can just touch all the files in the
DESTDIR before copying to /.


More information about the alfs-discuss mailing list