dev builds
themes manual
device status forums
mailing lists
IRC bugs
dev guide

Rockbox mail archive

Subject: Re: BUGS - ARGH


From: Daniel Stenberg <>
Date: Wed, 17 Dec 2003 08:54:26 +0100 (CET)

On Wed, 17 Dec 2003, BlueChip wrote:

> >If you can build it yourself, why use the 2.1 release code?

> Do you really need/want an answer to the above question?

Even if you can't use CVS for whatever reason, you can still get a daily
tarball that has newer code than 2.1.

> > > 4 warnings during compile:
> >
> >These warnings showed up in the build table for months before we fixed them,
> >*after* the 2.1 release...
> I didn't realise that the official release of v2.1 had known bugs that
> stopped a successful compile. How ironic (qv. 1st question)

Compiler warnings don't stop "a successful compile". They are just warnings.

> >Hm. Why do you get problems with this when others have not seen them? I'm
> >just curious so that fixing this with your proposal don't break it for
> >others.
> Can you name users that have seen this feature working correctly so that I
> may discuss it with them?

Back off.

This is a mailing list, we can speak openly about anything and others who read
this and disagree or agree about things we mention can speak up for

No, I don't generally write down the names of people that do things or that
say things that I afterwards believe I remember. I may even be mistaking.

Can't you at all understand my hesitation here? This code has been like this
since the very first win32 simulator (brought by edx). It just have to have
worked for some people at some point, and I want to understand why it works
for some and don't work for some before we go ahead and change it as you

> >When you run the tools/configure it creates a define in the makefile for
> >what target model you build for. In the source code, the
> >firmware/export/config.h is included and that checks what model you build
> >for and include the proper config-*.h header file, in which those defines
> >are defined.
> Yes, I thought that too - but it doesn't ...for either me or earHertz. I
> wonder why it works for you?

I'd rather reverse the question, why doesn't it work for you? What parts of my
description above fails for you?

> Can/Will you supply more accurate information about your development
> environment?

I don't think that is necessary to track down the problem you experience. Lots
of people build Rockbox just fine, on both windows and Linux/unix.

I build Rockbox for all models/targets on several different linux machines,
using different versions of the sh-elf compilers as well as the native

> > > main_menu options Plugins and Firmware do not work that is, selecting
> > > those options does nothing, it just leaves you on the main_menu screen
> > > this means that plugins cannot be used or tested in the simulator
> >
> >No clue.
> Which O.S. did you test it on?

I haven't tested that in a long while, but I only develop Rockbox (and run the
simulator) on Linux.

The win32 version often lags behind in functionality, so perhaps there's a
flaw in there.

> >Make a patch:
> >
> >
> "diff -u oldfile newfile > patch
> OR
> diff -u olddir newdir > patch"
> ????? ...You don't know what type of bullets I use ;)

I'm sorry, but I don't understand. Why is the above quote not a good
description of how to make a patch? What else do we need to explain?

> >-N "treat absent files as empty"
> bingo! diff file which includes objects, makefiles, .deps, .rocks, etc.
> etc. etc.

Yes, you can do that if you want. You can also exclude files, or just edit the
patch file after having generated it in your favourite text editor.

> Anyway, I am sure I can work it out one way or another. As you said at the
> top, it is good to repeat other peoples work, it makes you bored :)

That's not what I said.

 Daniel Stenberg -- --
Received on 2003-12-17

Page template was last modified "Tue Sep 7 00:00:02 2021" The Rockbox Crew -- Privacy Policy