On Monday, April 20, 2015 at 9:14:23 AM UTC+5:30, Chris Angelico wrote: > I definitely don't see how a non-text source code format would improve > on it. Feel like elaborating?
You are putting emphasis on the 'non'. This puts you into an oscillatory system between tautology and contradiction: How can something which is NOT be something? -- contradiction It is something else -- tautology If you dont put the emphasis on the not but on 'structured text' it may be more meaningful; eg html html is not 'not-text' -- you can edit it in a text editor alright. However if you edit it in a html-editor -- mozilla composer, seamonkey, or any of zillion web versions -- you get (at least) two views -- text vs structured In the text (ironically also called html) view it behaves like a half-assed text-editor In the structured view it renders the html structure (somewhat) So a table does not show as <table></table> <tr><td> etc but actually as cells. This prevents trivial counting/off-by-one errors. After the gross-layout is taken care of you can switch to text mode and add fancy html attributes if needed. Likewise having ready access to the AST of a program would be neat So for example when I used to do a lot of lisp, my editor was set up so that - cursor arrows behaved like normal cursor arrows - keypad arrows moved up/down/along S-exp ie they were tree movements -- https://mail.python.org/mailman/listinfo/python-list