--- Quote Start ---
Ugh. Sorry, but not an easy one. In justification:
- It is difficult to write IP that has no warnings, but parameterized IP is significantly more difficult. (It's just too easy to let synthesis reduce a bus to the correct size of something like that).
- It is difficult to say a warning is all right under all conditions. We don't want to do a blanket suppression, as there may be cases where the user has hooked something up incorrectly, and the warnings in the IP are what would identify that. Although with too many warnings, it's probably too difficult to know what's real and what's not.
Anyway, it's difficult, but Altera should do a better job of it, no doubt about it. I know this is always on the radar, but not sure what/when will happen.
--- Quote End ---
I'd say as an apology, rather? Relying on the synthesizer to reduce a bus to the correct size and not bother about the warning is pure laziness. Do not forget that the poor user has to search the messages for those warnings he's responsible for. And I always ponder whether those Altera warnings are benign or not, and whether it is me triggering them. Remember the old C-adagio:
a warning is an error waiting to happen. And personally I don't think it is difficult to write warning free code (even parametrized) , it may be a bit more work though.
Blanket suppression of Altera (and others!) IP generated warnings may be not that good idea, but you don't expect me to wade through all 1000 warnings on every compile?
An idea: maybe Altera can create a 'user warning' class with associated colour for warnings triggered by 'user' code?