From: Oren Ben-K. <or...@ri...> - 2001-12-12 02:45:57
|
Brian Ingerson [mailto:in...@tt...] wrote: > What *I* was proposing was a _shortcut_. Merely eliminating > some excess > whitespace. And it acheived Oren's goal, but was not limited to his > goal. In addition, this can be delayed under our current spec AFAICS. Right on both counts. > What *you* are proposing is a _change_ in syntax, with new indicators, > etc. I'm not willing to go there. I agree. Now, speaking of new indicators... I thought about the "format" business some more. The way I see it we have two self-consistent options. Either we go for broke and define a new indicator for format (% leaps to mind): "!type %format", and accept that sometime just the format may be specified without a type (a problematic issue); Or we decide that type and format are both parts of the same "transfer" property and use "!type!format". This way a format may only be specified if there is a type. I dislike ':' - it doesn't look good and is an awkward half-breed between a separate property and a part of the transfer property. I think that using "!type!format" is our best bet. It doesn't introduce a new indicator, and it does everything we need. Thoughts? Oren Ben-Kiki |