>A final question which has long interested me and which seems
>relevant to this whole discussion: who would use Tcl if it DIDN'T
>have such a integrated interface to Tk?
We used Tcl to access a database of statistics distilled from Web
server logs, and produce web pages showing summaries and graphs.
Many posters have touted Scheme over Tcl, on the basis of the design
of the language. However, in my situation, semantics were only one
consideration out of many. Perhaps my case is common, so I'll
describe the factors we weighed:
- Is the interpreter easy to embed in an application? Can the
application and the interpreter easily exchange data?
- Is it well-documented?
- Is it buggy?
- Can it talk to the outside word (in our case, Unix) well?
- Has it been ported to my machine?
- How big/slow is it?
Tcl scored adequately on all counts, and its semantics were good
enough. I don't know of any Scheme system that does (yet).
Perhaps what's missing for Scheme's success is simply sufficient
attention to annoying practical details, and not perfection of design.
Put another way, maybe we need to define the "design" to include the
entire development and use process, not just the language's semantics.
If this is so, SCSH and (I hope) Guile are going in the right
direction.
|