Re: [PATCH 10/15] Class regcache_readonly

[Simon Marchi <simon dot marchi at polymtl dot ca>]

On 2018-01-24 12:37, Yao Qi wrote:
> On Wed, Jan 24, 2018 at 4:57 PM, Simon Marchi 
> <simon.marchi@ericsson.com> wrote:
> > Just pitching some ideas, I don't think I understand the situation as 
> > well as
> > you do.
> >
> > I assume we want to keep the "regcache" type to mean read/write and 
> > attached,
> > since that's the most common use case.  Keeping this will reduce the 
> > amount of
> > changes needed throughout the code base.  We can then qualify the 
> > other types
>
> Yes.
>
> > based on how they differ from "read/write" and "attached".  That would 
> > give us
> > (in the same order as your list above):
> >
> > - readonly_detached_regcache
> > - detached_regcache
> > - regcache
> > - readonly_regcache
>
> This should be readonly_attached_regcache.

The logic was that by "default" a regcache would be attached, unless 
specified otherwise (because that's what the plain regcache is).  So 
that's why I suggested qualifying the detached regcache as such, while 
the attached ones would be implicit.

> > This would give a predictable naming, and makes it maybe easier to 
> > know what
> > to expect from each type.  The graph you used in message 0/15 would 
> > become:
> >
> >                       reg_buffer
> >                          ^
> >                          |
> >                    ------+-----
> >                    ^
> >                    |
> >             readable_regcache (abstract)
> >                  ^
> >                  |
> >            ------+------
> >            ^           ^
> >            |           |
> >     detached_regcache readonly_detached_regcache
> >           ^
> >           |
> >       regcache
>
> This naming is fine to me except for readonly_detached_regcache
> as it is too long.  As "readonly" implies "detached" in current 
> context,
> can we name it readonly_regcache?

In this context readonly == detached, but aren't we going to want 
readonly attached regcaches at some point?  If so, there will be a 
clash.

Simon