Re: [avr-chat] The silent death of -mint8

[Ruud Vlaming]

On Friday 01 July 2011, you wrote:

> This is nothing new. The developers are aware of this, 
> and we've been talking about it recently. 
Well, i am walking in the dark then. Although i am a occasional avr developer
(only when a project comes allong) i monitor four avr lists and avr freaks, and
saw nothing about this subject passing by. Where did i miss the buzz?

> -mint8 was already a hack, it never worked with avr-libc, and if you 
> ever ran into a bug with it, you've always been on your own. 
A lot of routines inside avr-libc work perfectly with that switch, and
i don't know about others, but i mainly use avr-libc for the defines,
no so much for the code. Especially not when the size is limited.
Btw, embedded programmers (should?) never use lib's blindly, and
always try to understand what is going on 'under the hood'.
 
> Yes, we all know that the point is to improve the optimization such 
> that -mint8 would never be necessary. I know that Georg-Johann 
> Lay has been working on that recently in trunk.
Nice!

> And it's wrong of you to think that we "forget" our developers, 
> or that we have no interest in improving the situation. 
With all due respect, and this is not a hollow phrase, because i really
do respect your great work for the avr gcc community, what i
think is entirely my business. 

But if you reread the concern i stated, you will see that i never 
said that 'they' would not be interested in improving the situation. 
However, by just dropping mint8 support, it was for example not 
mentioned in the changes:
   http://gcc.gnu.org/gcc-4.5/changes.html
they forgot the developers who made use of that switch. It is as
simple as that.

If you want to "bash" me, thats fine, but i kindly request you to stick
to the facts, and not react to something i did not say.

> Why not try your hand at helping to fix the avr backend of gcc?
Now that my friend, would be nice, but is not very realistic. I tell you why.
The last one and half year i made intensive use of many opensoure
projects. for example VLC, FreeCad, OpenFoam, Lazarus, several JavaLibs,
avr-gcc and probably  even many more without knowing. In most of them 
i encounterd deficiencies. Surly you realize that i use these projects to
get some work done, not just for the fun of it. So, helping them all out
is not an option. And if i would start working on avr-gcc why not on any
of the others? There is simply not enough time. You have to pick 
your battles.

I have an open source project of my own. In my view it works best if 
developers choose for one project and put all the effort there. So i try 
listen to the concers of my users, and try to keep the project as bugfree 
as is reasonably possible in my spare time.

With respect to avr-gcc i see myself as a user who signals the problems
i encounter to the list, (and occasionally give advise on the list if i know
the answer straight away), but i leave it up to the developers to act or not.
I wish it could be otherwise, but it isn't so we have to live with that.

Ruud