3.20X vs. 3.191

[valemike]

I've been using 3.191 since it's a baseline release. When I got to 100% ROM usage, I figure i'd give 3.207 a try and downloaded it yesterday. I haven't really done much of a regression test, but it pretty much all works.

On a PIC18F252, I noticed that my earlier 100% ROM usage got reduced down to 80%!

Is this what is expected with 3.20X, i.e. a 20% reduction in 18F generated code? That's pretty good.

[Jeff King]

We went from 3.184 to 3.207 and got about the same reduction in code size, and other then the structure bit test thing (which apparently got fixed in 3.208) everything is working OK. This is a tcp/ip and 802.11b stack, on a 18F8621, so it is fairly complicated and works out alot of the code.

None-the-less, I do wish CCS would test things more extensively and post a list of know bugs on each release as they are discovered. In addition, it might be useful to leave older releases online (like for example 3.207 which was replaced by 3.208) in case the "new release" breaks the old release. Some bugs you can work around, others you can't, but digging through user posts is not my preferred way to document them. Ideally, as customers/CCS discovers them, CCS can confirm them and repost the bug to a master list, referernced by version.

In other words, CCS needs to put itself in the shoes of a company shipping a product, and less in the shoes of a hobbyist, although I think you can have it both ways.

-Jeff

p.s. 3.207 also seems like it compiles way faster then 3.184