Should "Adventure Game Studio" be a tool primarily aimed for making games?
Or should AGS primary goal be to run older games and having the whole "making games" as a secondary feature?
Why would AGS primary goal be to run older games? In fact, similar to Snarky's point, why should older games be a priority at all?
This is a case of manipulation of public opinion, because the question was never put how Alan worded it.
"Making games" was never suggested as a "secondary feature".
Running old games was suggested as
one of the integral features of the
engine in the past, since JJS started making official ports in order to run AGS games on other systems, because prior to that it was near to impossible to reliably run them on other systems than Windows, and even run games compiled many years ago on contemporary Windows, - that also was becoming more difficult (old executables did not have support for modern graphic modes, for example). AGS database had many hungreds of games at that point, and most of them were running only Windows. Users on Linux and Mac desired to be able to play them natively without virtual machines and such.
EDIT: In case someone wonders how much used this feature is; this may be not seen on this forums, but in all these years I was receiving support requests from dozens of people, who wanted to run pretty old games on modern devices. There are still people who add translations for old games too, and being able to run them with new engine for them is a big deal.
The question was not whether this should be primary feature or not, the question in a nutshell is whether it may be discontinued for the benefit of a major redevelopment.
From the start of the current open source project it is based on JJS's works, which included layers of backwards compatibility (some of that was not a part of 3.2.1, last version by CJ). Unfortunately, the engine code was never cleaned up enough to untangle this compatibility support and make it an optional module, for instance. Because of that it was always difficult to make bigger changes, while I was reluctant to drop it thinking that we are now responsible of carrying this support, at least until there are better alternatives.
Personally, I believe that was a big mistake, and would've things done differently, AGS probably will be pretty more advanced than it is now. Of course that would be achieved at the cost of people not being able to run old games with this engine, but perhaps same functionality could've been implemented in another, parallel, project, or even in ScummVM.
And yes, it's true that this was all discussed many times before.