Build for Classic

[Richard Gaskin]

Robert Brenstein wrote:
> It is not only about new features. What about having to keep two 
> different versions of Rev as a developer?

Indeed, if they spent tens of thousands of dollars a handful of their 
customers could enjoy the relatively minor convenience of clicking one 
Build button rather than two.

But think about it:  Even if they gave us a Classic engine today, the 
feature parity would be short-lived.  Soon after there would be a new 
version of Rev, and we all understand they can't support forever an OS 
Apple themselves abandoned long ago.

Within weeks of any new Classic engine the parity convenience would be 
gone as the engine continues to move forward.


> RR kept saying that newer version for OS9 is coming, for whatever 
> that is worth, but haven't delivered. Just like you want RR to fix 
> your favorite bugs, Linux users want to get Linux version up to date, 
> people who are stuck with OS9 want to get a newer version.

I think the order you listed them makes good business sense. 
Fortunately that also seems to be the order in which they're being 
addressed.

> The fact that my favorite bug does not affect you can't be a reason
> for RR ignoring it, and this is what some people seem to be saying.

Stephen Barncard raised an interesting, perhaps definitive point:

   Branching would require Rev to use an old, outdated compiler.

I believe he's right:  AFAIK, there's no compiler which will let you 
build for both Intel Macs and Classic.  All modern compilers have 
abandoned Classic.  This would mean that not only would building for 
Classic require forking throughout the codebase, but would require 
forking every element in their process, and maintaining a separate set 
of outdated headers, some of which may be incompatible with modern 
replacements and require additional forking.

In short, an extremely costly mess.

And all to deliver 15 minutes of feature parity for people who can't 
really use most v2.7 and 2.8's new features in Classic anyway.

I hadn't thought about the compiler angle before, but if it's as hairy 
as I guess it would REALLY benefit RunRev to just have a frank 
discussion here about the disproportionate costs, encourage people to 
use v2.6.1 for Classic builds, and move on with their busy day.

> Ending full OS9 support would be easier if RR did proper branching of 
> their releases, so they could continue to fix critical bugs in 
> whatever is the last OS9 release.

We're currently seven years past Apple's kill date for OS 9.  Apple 
themselves no longer provides any patches for it, and haven't for more 
than half a decade.

Exactly how long do you feel it would be reasonable for a third-party 
vendor to exceed Apple's commitment to the OS they abandoned?

-- 
  Richard Gaskin
  Managing Editor, revJournal
  _______________________________________________________
  Rev tips, tutorials and more: http://www.revJournal.com