Re: Further problems with makeLSR

[Phil Holmes]

----- Original Message ----- 
From: "David Kastrup" <address@hidden>
To: "Phil Holmes" <address@hidden>
Cc: <address@hidden>
Sent: Friday, November 02, 2012 4:21 PM
Subject: Re: Further problems with makeLSR


> "Phil Holmes" <address@hidden> writes:
>
> > ----- Original Message ----- 
> > From: "David Kastrup" <address@hidden>
> > To: "Phil Holmes" <address@hidden>
> > Cc: <address@hidden>
> > Sent: Friday, November 02, 2012 3:41 PM
> > Subject: Re: Further problems with makeLSR
> >
> >
> > > "Phil Holmes" <address@hidden> writes:
> > >
> > > > From: "David Kastrup" <address@hidden>
> > > > To: <address@hidden>
> > > > Sent: Friday, November 02, 2012 3:24 PM
> > > > Subject: Re: Further problems with makeLSR
> > > >
> > > >
> > > > > Oh rats.  The problem would be that any changed snippets need to be
> > > > > copied to snippets/new (after editing their headers appropriately) in
> > > > > order not to be overwritten by LSR.
> > > > >
> > > > > And, of course, after the latest change this concerns a sizable number
> > > > > of snippets.  We need to get this done before the next LSR update.
> > > > > Anybody up for it?
> > > >
> > > > I think that _might_ not be necessary.  If it's possible to update
> > > > them with a convert-ly rule, they should not need adding to
> > > > snippets/new.
> > >
> > > Obviously that does not help since all of the affected snippets were
> > > actually changed with convert-ly.
> >
> > Is that rule definitely in master?
>
> It definitely is.  Iff the convert-ly target is 2.17.6 or greater, the
> rule should be triggered.
>
> > MakeLSR runs convert-ly and therefore should come up with the same
> > change that you produced.
>
> So we need to figure out why it doesn't.  Does it run the wrong version
> of convert-ly?

It says it's running 2.17.6 - although that could cover a multitude of 
revisions.  I'm using latest git master.  Is the rule definitely in master?

> > > > Otherwise, we'll end up with too many snippets in /new to be
> > > > comfortable with.
> > >
> > > Where does the comfort level derive from?
> >
> > Human beings have to move all the snippets from snippets/new once they
> > work with the current version of the LSR.  A few 10s is achievable.
> > More than that risks no-one having the incentive to do the drudge.
>
> But they won't work with versions of the LSR previous to 2.17.6.

They don't need to. Snippets in the LSR have to work with the current 
version of lily that the LSR runs - currently 2.14, IIRC.  If they can be 
updated by convert-ly, there's no need for them in snippets/new. 
snippets/new is really there for snippets that don't work at all and can't 
be converted - for example, one that uses a tick as a breath mark could not 
be placed into the LSR at all.

> > > > Alternatively - does the older syntax still work?
> > >
> > > _Some_ of the older syntax continues to work (namely that for
> > > \override/\revert).
> > >
> > > But I don't see that presenting an inconsistent view would make any
> > > sense here.
> >
> > I don't see why the snippets need consistency.
>
> I do.  We don't want to confuse users by presenting two different
> versions of syntax.
>
> --
> David Kastrup

--
Phil Holmes