Dear Robert,
I'm not sure I like your latest restriction of system names to what logical pathnames accept: one of the big pluses of not using logical pathnames everywhere was to allow for a wider range of system names. Consider the widely used cl+ssl, for instance.
At ITA, we also used to have module names like foo-V1.200, that are not compatible at all with your plan of forcing logical pathname compatibility — even less if converted to package-system where that would become part of the name.
Logical pathnames are a horror that no one should use, and the portability constraints of which should not be inflicted on everyone.
Could you revert this new constraint?
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org You can only find happiness by striving towards something else.
On Wed, Feb 26, 2014 at 10:14 AM, Robert P. Goldman rpgoldman@sift.info wrote:
This is a repair of last night's commit. I added a test for the new system name checks, which had some fan-out -- I needed to fix DEF-TEST-SYSTEM to permit systems named with strings instead of symbols.
Also, I check to make sure that all the system names, when being registered, are case-flattened down, and issue a continuable error if not.
cheers, r