Yep - that got me pointed in the right direction. Discovered I don't even need to technically "load" the different versions, I can just replace files under hardware/Cosa, force a VM rescan and replicate the error on demand.
With this, I narrowed down the single cause to the boards.txt file - I can install 1.2, replace that single file with the 1.1.1 version and compile without error.
Analyzing the file changes (apart from the significant number of boards deleted), there was only one difference...
v1.1.1
uno.build.core=cosa
v1.2
uno.build.core=Cosa:cosa
I had a (somewhat blind) hunt around the rest of the source and online to try to figure out what the significance of the colon might be, but couldn't find anything that mentioned it.
Had a quick look back at the original thread I found, and based on
http://www.visualmicro.com/forums/YaBB.pl?num=1421322718/11#11 I'm going to guess that broke your automatic detection of Cosa core boards
So, I just did a search/replace on the 1.2 boards.txt file, replacing Cosa:cosa with cosa, and success. I was able to compile, upload and hit the breakpoint without any problems.
Now with that said, I couldn't get @ReportDigital working - either with v1.1.1 or v1.2. The original thread said you were having issues, but I was hoping there'd been progress over the last year
Is there anything we/I can do to help? Is it still related to the idea that the pins are hardcoded constants?
On a side note, I'm getting intellisense errors on macro defined functions (such as PSTR). Intellisense correctly identifies the related #define command, but throws a "expected an expression" error...and I'm not sure how to (or if it's possible to) manipulate the header file to counter this
(as per Extending-Visual-Studio-Intellisense.aspx).
Thanks for your speed response so far, much appreciated. A pro license is already pretty much a foregone conclusion at this point. If I can get the pin reads working, it's guaranteed