Damien Cassou-2 wrote > Sven Van Caekenberghe < > sven@
> > writes: > >> I am not so sure we should add that. >> >> The JSON spec explicitly does not allow comments because of fear of abuse >> (that the comments would be used to add semantic meaning outside the >> spec). > > > really? That's surprising. Comments in configuration files are very > important. Even more when the configuration files are templates for new > users. I think you may have combined two requirements and come up with a "I can adapt this existing idea to handle both" solution. In my experience, I would suggest the word "adapt" in that phrase /usually ends up/ being better read as "pervert". So, let's step back for a moment and start by clearly and precisely identifying the actual requirements. As I interpret the foregoing discussion, there are two. 1) We want the ability to include information in a configuration which can be used to document, explain, guide those who come after. 2) We want to provide a template configuration that will make it easier for people to figure out how to create a good configuration. I will also remind everyone of one thing that seems to be a fundamentally important aspect of Smalltalk programming: we prefer code over comments. For example, I came across some unit tests which included comments like "The result should be blah". Generally, we would write that as "self assert: result equals: blah". So, let's add this requirement: 3) The best implementation of a "spec" like a configuration will maximize the DSL capability of Smalltalk. In my opinion, none of these requirements are well satisfied by adding comments to the STON syntax. The first requirement, again in my opinion, is best handled by modelling a narrative aspect in configurations. I'm not suggestion how that should be done nor what to call it. The second requirement seems best satisfied by providing a template configuration file with the its various aspects specified by some example content that clearly illustrates the right way to do it. And the third requirement is satisfied when the modelling of the configuration formalizes what it is we want it to convey. > -- > Damien Cassou > http://damiencassou.seasidehosting.st > > "Success is the ability to go from one failure to another without > losing enthusiasm." --Winston Churchill -- View this message in context: http://forum.world.st/Comments-in-STON-tp4830944p4830990.html Sent from the Pharo Smalltalk Developers mailing list archive at Nabble.com.
