Question on alternateGroup

[nickray]

#506 introduced "alternateGroup support".

In the case of LPC55 (e.g. https://github.com/lpc55/lpc55-pac/blob/d53269c5e87ae5151f780203702b548584dd0610/svd/pack-13.0.0-LPC55S69_cm33_core0.xml) this completely changed the svd2rust API all over the place.

I'm not sure if this was intended/foreseen?

Also not sure if this is a bug of sorts in the SVD; the net effect is it makes a lot of names very long (which I'm tending towards reverting by simply sed-ing out all the alternateGroup lines).

[burrbull]

Looking at SVD you cite as an example I see incorrect usage of alternateGroup.
For example SCT0 peripheral.
CAP14 and MATCH14 both at address 0x138. And both in one CAP_MATCH alternateGroup.
If alternateGroup is used they must be in different groups. All CAPx in CAP group and all MATCHx in MATCH group.
So LPC SVDs are bugged. But in this particular example alternateGroup is not needed at all.

The only way to workaround this is to add flag that disables #506 for such files.
It can also be partially fixed if we will check for same register names. And don't use alternate_group_ prefix if we don't find such registers.

[burrbull]

See #516

[burrbull]

You can now use --ignore_groups for old behavior.

[nickray]

Cool, thanks!