On Nov 26 22:01, Achim Gratz wrote:
> Ken Brown writes:
> >> + if (file.size() == file.rfind('/'))
> > ^^^^^^^^^^^^^^^
> > Shouldn't this be file.size() - 1?
>
> Yes, but it's anyway not the best code to check for a trailing '/'.Feel free to add methods which match your use case better. > > Is this last test necessary? The Script constructor guarantees that > > the stratum is allowable. > > You've convinced me that this safety check is no longer necessary (it > was with the initial version of the constructor, though). > > > 1. You should do something to make sure that incremental rebase (or > > whatever version of autorebase gets chosen) is run first among the > > perpetual postinstall scripts in stratum 0. One possibility would be > > to also allow stratum 1, and require that no packages other than > > incremental rebase use stratum 0. Alternatively, give the incremental > > rebase script a name that guarantees that it comes first in whatever > > order the perpetual scripts are run in. > > The perpetual scripts are different in that dependency order can't be > generated for them. ...except you allow to define more than 1 stratum and run the perpetual scripts stratumized. I thought that was the underlying idea. > The order in which they are run is the sort order > of the file find/visitor that picks them up from the postinstall > directory. That order may be locale dependent, so I think we'd somehow > need to force something sane here (POSIX). The order is the order on disk, which is using the LOCALE_INVARIANT locale when creating the B-tree, IIUC. But it's not asked too much if the filename character set is restricted to ASCII. Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat
pgpfRNi1qF4jH.pgp
Description: PGP signature
