Re: debug explain part 2

On 12/10/18 8:27 AM, Mats Wichmann wrote:
> this is part 2 of the debug=explain discussion.

yes, replying to my own message.  had a thought on one piece:

> + src/hello.c updated:
> scons: rebuilding `build/hello.o' because the dependency order changed:
>                old: ['src/hello.c', 'src/hello.h', '/bin/gcc']
>                new: ['src/hello.c', 'src/hello.h', '/bin/gcc']
> 
> This is because a later check uses whether or not any explanation lines
> have been added yet as a flag in front of checking something else:
> 
>         if len(lines) == 0 and old_bkids != new_bkids:
>             lines.append("the dependency order changed:\n" +
> 
> and because we already know one element of each list is the object for
> src/hello.c, which differ, then the two lists cannot compare equal.

old_bkids and new_bkids are the combination of three lists: sources,
depends and implicit.  Should the check for "dependency order changed"
actually include the source(s)?