Rockbox.org home
release
dev builds
extras
themes manual
wiki
device status forums
mailing lists
IRC bugs
patches
dev guide



Rockbox mail archive

Subject: Re: HWCODEC

Re: HWCODEC

From: Dominik Riebeling <dominik.riebeling_at_gmail.com>
Date: Sat, 24 Dec 2011 16:46:11 +0100

On Sat, Dec 24, 2011 at 2:07 PM, Thomas Martitz <kugel_at_rockbox.org> wrote:
> I'm also an advocate of keeping stuff similar wherever possible. This will
> actually simplify code, and also give it more testing (because it's running
> on more targets) so bugs are (hopefully) found earlier.
>
> You advocate, fine. But you don't _do_. Seldomly, at least. I acknowledge
> the recent talk.c refactoring is positive exception.

How often have others by making / unifying things? Why is it a
"positive exeption" if someone who is mainly inactive these days
starts working on some stuff again? I find such a "positive exception"
statement pretty inappropriate.

> This has been done. But the calls are often unheard. You're right about the

Maybe it's also because those calls are placed in the wrong channel?
Do you really expect inactive developers to monitor the project all
the time? I don't consider myself inactive and am not able to monitor
all IRC chat that's going on. IRC is really a nice medium but it's
rather bad if you're not able to be around and follow its discussions
regularly. Since people tend to have other obligations (like a day job
for example) this can be quite a serious issue (at least for me it was
*much* easier following Rockbox development when I was a student than
these days having a fulltime job, and it was *much* easier to actually
find the time to get things done).

> It's both. It's a problem of too few *active* developers testing/running
> stuff on HWCDEOC. And it's a problem because Rockbox has so many features,
> they just can't be covered by the very few HWCODEC devs.

which pretty much means something I'd like to see since quite a while:
automated (unit-)tests. And it's not a HWCODEC problem, the same
problem exists on SWCODEC. The only difference is that there are way
more users for SWCODEC targets, and (at least some of those users)
file bug reports. There are rarely used feature that can have bugs for
long before anyone reports them, and that's true for SWCODEC as it is
for HWCODEC. In fact, since Rockbox has much more features on SWCODEC
I expect may more hidden bugs in SWCODEC code than in HWCODEC code.

The number of features Rockbox has these days makes it simply
impossible for any dev to test everything. There was some discussion
about automated test like one year ago IIRC. And also IIRC people did
agree that automated test would be a good thing. Nothing happened
since. Where have the active devs been since? Why did nobody work on
automated (unit-)tests? Why don't we have any test programs for e.g.
buflib in svn? You surely used some test programs during development,
didn't you?

> Also code unification is nice, but it doesn't solve all problems and it
> doesn't make testing unecessary. Many things can't be reasonably implemented
> in a unified manner. And even then, things cannot be tested properly.

Why can't unified things be tested properly? Unified stuff is actually
*way* better for testing, since the target dependent stuff can be
replaced with test stubs or an implementation on a host machine.

 - Dominik
Received on 2011-12-24


Page was last modified "Jan 10 2012" The Rockbox Crew
aaa