Re: [PATCH v4 6/7] disas: mips: Add micromips R6 disassembler - infrastructure and 16-bit instructions

[Aleksandar Markovic]

On Saturday, January 25, 2020, Richard Henderson <address@hidden> wrote:
> On 1/24/20 1:38 PM, Aleksandar Markovic wrote:
> > On Friday, January 24, 2020, Richard Henderson <address@hidden
> > <mailto:address@hidden>> wrote:
> >     The thing I'm concerned about here is any future maintenance of this file.  One
> >     would surely prefer to edit the original decodetree input than this output.
> > 
> > 
> > Here is the deal: This dissasembler is meant to reflect the  documentation of a
> > particular ISA, and as the documentation largely stays constant (except adding
> > some insignificant errata), the disassembler will stay virtually constant, we
> > wouldn't like even to touch it, and we like it this way.
>
> No, this is neither right nor proper.
>
> To review the code in this form is significantly harder than in its decodetree
> form.  That is in fact the whole point of the decodetree form: otherwise we'd
> still be writing these sorts of parsers by hand.
>
> While there's no license on this new file (another problem), if as assumed this
> is GPL 2+, then you are in violation of the GPL.  From section 3:
>
>   # The source code for a work means the preferred form of
>   # the work for making modifications to it.
>
> You cannot with a straight face claim that the generated c is the preferred
> form for making modifications.
>
> Finally, suppose we improve decodetree.py, so that it produces code with which
> the compiler produces better code, for some metric of better.  We would want
> this disassembler to benefit as well.

I think you got some things upside-down. A tool developent and usage should be driven by the needs of users of a tool, and not dictated by the author. Users should be free to use the tool in any way they seem suitable, including its modification.

> r~