[Rubygems-developers] PATCH: respect requirements of gems

[Mauricio Fernández]

On Fri, Apr 23, 2004 at 07:33:26AM -0400, Chad Fowler wrote:
> >We fully agree on this; I just see the API as a part of the dependency.
> >
> 
> I do too.  And, I think everyone who knows anything about programming 
> would too.  But, I don't think we have to enforce everything in code, 
> in the same way that I don't think we have to check "type" for method 
> parameters, or make everyone follow the same naming conventions with 
> code.  I believe the inclusion of Jim's additional 1.*.* kind of thing 
> will give us the functionality we need to *recommend* a sensible 

What's wrong about requiring a specific internal version numbering
scheme? I mean, the gemspec already requires rubygems_version, name,
platform, date, summary, require_paths, version; what's so wrong about
adding api_version to the list? Keep in mind that people would still be
free to use whatever scheme they like for the 'main' version number.

> convention for version numbering of gems.  And, the community in actual 
> practice will adhere to it.  There may be a bump or two occasionally, 
> just like occasionally someone changes an API without warning everyone 
> and breaks code or makes a change in CVS that has unintended 
> consequences.  It's then caught and fixed.  We can't write code to 
> guard against every eventuality.  We can write enough code and give 
> users a framework so that it's *possible* to do the right thing, but 
> I'm very much against trying to enforce policy like this via code.  I 
> believe Jim's idea is what is needed to move us in the right direction 
> w.r.t versioning.

Actually, Jim's idea applied to the version number imposes far more on
the lib. devel than the internal API versioning: if gemmakers start
using 1.* that'd mean that libs. are expected to change their major
version num. on API change. The code I sent requires this for the
_internal_ version number, so the 'main' one can be set to whatever the
devel. wishes.

> 
> My feeling is that this isn't nearly as big a deal as it's being made 
> out to be.

What are the main pending issues in your opinion then? (not a rhetorical
question, I really want to know).


-- 
Running Debian GNU/Linux Sid (unstable)
batsman dot geo at yahoo dot com

"You, sir, are nothing but a pathetically lame salesdroid!
I fart in your general direction!"
	-- Randseed on #Linux