Re: Hardcoded LP version in *2ly scripts?

[Colin Campbell]

On Mon, 2010-12-27 at 23:14 +0000, Graham Percival wrote:
> On Thu, Dec 23, 2010 at 07:30:55AM -0700, Colin Campbell wrote:
> > 
> > Convert-ly was the one which got me into tracing the issue; it looks as
> > though *all* the related *2ly routines have the same code,
> 
> Yes, this is correct.
> 
> > and all are hardwired to a specific version.
> 
> Yes, this is correct and done in the GUB building.
> 

Just for curiosity, would it be feasible to code them to use the latest
version of lilypond installed in /usr? IOW, is there enough change
between LP versions to create such a tight dependence as to require
hard-coding?

> 
> > > PS virtually nobody runs "make install", and I certainly wouldn't
> > > expect a new contributor to do this.
> > 
> > It seems easy enough to me; the instructions in the CG are clear, as to
> > building, anyway.
> 
> Yes, but obviously you got into trouble somehow.  :)
> 

In a PPOE, a development shop running MAPPER-1100, my job was to break
the code. We lived by the rule that code cannot be made fool-proof, only
fool-resistant.


> If anybody is seriously interested in the build system details,
> the source is there; me spending a few hours trying to explain it
> would be less useful than spending that time towards getting 2.14
> out.
> 

Someday, I'd love to find out how the recursive directory links come
about: I tried to grep -R 2ly over ~/lilypond-git and was taken quite
aback at how huffy grep gets about recursion loops!

> > 
> > Get well soon, Graham, and I hope your Christmas is a happy and restful
> > one!
> 
> I was sleeping for 18 hours a day, which was fairly restful.  I
> wouldn't call it "happy", but at least I'm well now!

Good news, Graham! Glad you're well on the mend.

Colin

-- 
In knowledge work, the first question is, 'What should you be doing?'
Not how. - Peter F. Drucker, 1909 - 2005