Re: Re: git: 50c28128f2ef - main - comms/lrzsz: do not link with libbe(3)

On Sat, Jan 27, 2024 at 05:09:47PM +0100, Mathieu Arnold wrote:
> On Sat, Jan 27, 2024 at 03:59:19PM +0000, Alexey Dokuchaev wrote:
> > On Sat, Jan 27, 2024 at 03:46:55PM +0000, Dirk Meyer wrote:
> > > commit 50c28128f2ef196caf67619027335a44b005641d
> > > 
> > >   comms/lrzsz: do not link with libbe(3)
> > > 
> > >   Submitted by: Dmitry Lukhtionov
> > >   PR: 276613
> > 
> > To make the attribution more transparent, the actual patch was
> > conceived by eugen@ (Eugene Grosbein).
> 
> If only it was actually a patch and not some sed stunt...
> 
> We really need to stop doing this, a patch or a sed will both break if
> the file gets modified, but a patch actually shows the intent of what is
> being removed and makes it easier to update, wheras here, it removes 140
> lines starting at some string.

While I'm a big sed(1) fan myself, recently I've been leaning more towards
what you say above, however: in this case, it's an old code, looking for
syslog in some obscure places like -lbe (BeOS library, not the modern one
for managing ZFS boot environments).  Correct patch should not be removing
140 lines at certain place, but fixing the search logic instead.  Until
someone(tm) could upstream it, sed(1) hack for some low-volatile code is
good enough (and with git, easier to manage vs. regular patches).

./danfe