[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [RFC] Modification history as a source code

On Tue, Jun 17, 2003 at 10:49:46PM +0100, Edmund GRIMLEY EVANS wrote:
> Dmitry Borodaenko <d.borodaenko@sam-solutions.net>:

> > even to its creator, 'form preferred for modification' should be chosen
> > from forms remaining in existence.

> You shouldn't be choosing at all. You should provide everything that
> is likely to be useful.

Where a minimum standard for free software is concerned, this is asking
a bit much.  The requirements, as I understand them, are that 1) the
provided source corresponds directly to the binaries being distributed,
and 2) it is the form of source code most suitable for editing (as
judged by whether it's the form the creator has actually used for
editing).  "Everything that is likely to be useful" imposes an undue
burden on the author to provide lots of things that, because they don't
meet the above two criteria, are non-essential.  There are plenty of
things that are likely to be useful in one context or another that we
shouldn't require the distribution of.

> For example, if you automatically converted a Pascal program to C,
> then made small changes to the C, then you should provide both the
> Pascal and the C as both are likely to be useful to at least some
> people who want to make changes.

No, you should only provide the C source, because the binaries being
distributed are those of the modified C program.  Once I've started
editing the C program, I've made it unambiguously clear that this is the
preferred form for modifications; just because the Pascal source exists
doesn't mean I should have to distribute it.

> On the other hand, if the C has been worked on a lot and nobody has
> used the Pascal for a few years, then there is probably no need to
> continue distributing the Pascal. In any case, you have to draw the
> line somewhere or old versions of the code will pile up like invariant
> sections in an FDL document ...

Exactly.  So draw the line where it needs drawing -- don't impose a
requirement to distribute files that the programmer isn't in fact using
to create the binaries.

Steve Langasek
postmodern programmer

Attachment: pgph_l2MpyRE3.pgp
Description: PGP signature

Reply to: