Re: CVS commit: src/sys/dev/ic

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

macallan@ wrote:

> On Thu, 12 Jan 2012 18:19:26 +0900
> Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost> wrote:
> 
> > > Module Name:      src
> > > Committed By:     macallan
> > > Date:             Wed Jan 11 20:41:28 UTC 2012
> > > 
> > > Modified Files:
> > >   src/sys/dev/ic: igsfb.c vga.c vga_raster.c
> > > 
> > > Log Message:
> > > wsfont_matches() and wsfont_find() take an extra parameter now
> > 
> > Isn't it possible to provide compatible wrapper functions (without
> > an extra parameter) rather than changing all existing API callers?
> > 
> > So that you don't have to bump kernel version for modular(7) and
> > all third parties (including ongoing porting efforts) don't
> > have to fix their drivers on updating code base.
> 
> That would be trivial to do. The reason I didn't do it right away
> was that I had no idea there were so many drivers that call
> wsfont_find() - at least that shouldn't be necessary anymore
> in most cases, now that rasops_init() at least tries to pick
> a sensible font for the screen / terminal size requested.
> In vga it makes sense ( in order to get an 8 pixels wide
> font and nothing else ), in igsfb not so much.

 - No discussion/review on public tech lists against public API changes

 - No man page update after API changes

 - It seems you didn't notice breakage even in MI vga.c on your first commit

 - evbarm and hpcarm (and zaurus) are still broken on daily build
   due to API changes

It looks your changes are a bit rude and violate our commit guidelines
unless they were approved by Core or other responsible persons.

> Also, I'm not aware of any modular wsdisplay drivers.

The kernel version number represents ABI compatibility
regardless of whether there are actual referers or not.

If you don't want to bump version, keep ABI compatibility
as you say it's trivial, or ask Core if your changes are okay.

Thanks,
---
Izumi Tsutsui