Re: pbulk: What can I expect from reuse_scan_results?

[Joerg Sonnenberger <joerg%bec.de@localhost>]

Am Wed, Apr 27, 2022 at 03:24:45PM +0200 schrieb Matthias Petermann:
> 
> Hello Jörg,
> 
> Am 26.04.2022 um 22:44 schrieb Joerg Sonnenberger:
> > Am Tue, Apr 26, 2022 at 02:47:28PM +0100 schrieb Jonathan Perkin:
> > > * On 2022-04-26 at 14:37 BST, Matthias Petermann wrote:
> > > 
> > > > However, it takes the same amount of time as without reuse, and the
> > > > output on the terminal is the same as the previous run. Is this normal
> > > > behavior, and if so - what can I generally expect from
> > > > reuse_scan_results?
> > > 
> > > It's normal.  Re-using scan results will improve performance if there aren't
> > > many changes, but for example devel/gmake was touched in the last few days
> > > and that will cause pretty much everything to be recomputed, and you won't
> > > see any difference in runtime.
> > 
> > Actually, touching a pure tool doesn't force a rescan, but touching
> > something in pkgsrc/mk/ like bsd.prefs.mk will certainly force
> > everything to be rescanned.
> > 
> > Joerg
> 
> Strictly speaking, I feel that the re-scan is also taking place even though
> I have not made any changes to the entire pkgsrc directory.
> 
> I use a limited_list and saw in /usr/pbulk/libexec/scan that there is a case
> distinction here - depending on whether the variable is set or not. Can this
> have something to do with it?

Oh, there is no real merge logic for partial scans, so no use for them.

Joerg