regression tests

[Graham Percival]

Jan Nieuwenhuizen wrote:
> Ok.  There it says
>
>    if there is any regtests which is useful as documentation and there
>    is no corresponding snippet in LSR
>
> Although I agree that would be nice, currently that statement is
> ridiculous.  It is way too early for that?  We have about 600
> regression tests, and there are only about 220 lsr snippets.

And only about 10% regtests are useful *as documentation*.  60 easily 
fits inside 220.  If we missed a few, we can add them.

Most of the regtests are easily covered by the manual.  Especially now 
that we're adding new regtests whenever we fix a bug (which is quite 
appropriate)

What's the documentation value of tie-grace.ly ?  It works exactly like 
you'd expect it to work.  Or beam-cross-staff-script.ly ?

>  I agree
> that newbies should not look at the regression test, but if you shut
> out power-users from the regression test, they'll miss about 400
> features?  We will have all sorts of questions: Can lily do this?

Power-users know how to read the program reference.  They can see the 
features there.

Look, the regression tests are not _intended_ as documentation, and they 
_should not_ be intended as documentation.  They are regression tests! 
Users should have three places to look for help:
- manual
- program reference (advanced users)
- LSR
- special-approved-lsr-snippets-in-our-docs

(ok, that's 3.5 places)


I wish that more users searched the mailist archives, but they don't. 
Useful tips sent to the mailist are essentially lost knowledge; that's 
why I've really been pushing LSR.

> It's also is an amazing advertisement for lily's features?

Advertising goes in the Examples or in the new "inspirational headword" 
examples that are planned in GDP.

>  How is
> a user supposed to report a bug or do a feature request without
> having access to the regression tests?

Users report bugs without reading the *manual*.  They're not going to 
check the regtests.

Cheers,
- Graham