16:55 <guardian> ddfreyne: in nanoc 4, will you have dependencies on items' attributes? instead of whole item?

16:56 <guardian> ddfreyne: e.g. archive page only depends on items' titles, you don't need to know whether compiled content changed, only if title or date changed

18:33 <ddfreyne> guardian: Hopefully, yes!

18:33 <ddfreyne> guardian: To be honest, I’m not sure if I will finish nanoc 4.

18:34 <ddfreyne> guardian: It is a *lot* of work to get nanoc 4 right, and I am not even sure that I can achieve what I want to achieve.

18:34 <ddfreyne> (For example, dependencies that are always 100% guaranteed to be right… that is not possible as long as there’s ERB+Haml and a Ruby Rules file.)

18:35 <ddfreyne> Considering that nanoc 3.x is already quite nice, I am not sure whether having nanoc 4 is really that useful

18:35 <ddfreyne> To be clear: nanoc 4 is not cancelled, but it needs to get things *truly* right. In its current state, it has many of the same problems as nanoc 3 has

18:36 <ddfreyne> As far as I can tell, the only approach that leads to where I want nanoc to be (fast, dependency tracking on attribute level, no recompilation unless necessary, …) would involve dropping support for ERB and Haml too.