home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!uwm.edu!biosci!agate!tweety.cs.berkeley.edu!larry
- From: larry@tweety.cs.berkeley.edu (Larry Rowe)
- Newsgroups: comp.lang.tcl
- Subject: Portable applications in Tcl/Tk
- Message-ID: <1er9ddINN6gb@agate.berkeley.edu>
- Date: 23 Nov 92 18:54:37 GMT
- Reply-To: larry@cs.Berkeley.EDU
- Organization: U.C. Berkeley EECS Dept.
- Lines: 72
- NNTP-Posting-Host: tweety.cs.berkeley.edu
-
- (Commercialization of Tcl/Tk discussion. Skip message if you will be )
- (offended by such discussion. /larry )
-
- Interest in Tcl/Tk has been booming. John says the USENET ratings estimate
- that over 14K people are reading the newsgroup. Of course, this estimate is
- probably high, but still I'm sure that everyone perceives that traffic is up.
-
- From a strategic computing perspective, it is clear that the development of
- future GUI applications *must* run on all platforms -- i.e., MacOS, Windows,
- and X. Currently, tk runs only on X. Several people have indicated an
- interest in a window system independent tk, call this TK*. Problem is that
- the task is probably too much for a volunteer programmer to complete in a
- reasonable amount of time.
-
- So, it seems to me that we need a commercial company that will provide a
- supported TK*. Problem is how to fund it. Venture capitalists might be
- interested in funding such a company, but they want something with the
- potential to grow to $50M-$100M. It's hard to predict how tcl/tk by itself
- will grow that large. On the other hand, the limiting factor to its growth
- may be the lack of commercial support and TK*.
-
- I've had experience building commercial application development tools.
- I've also given thought to trying to do the tcl/tk development system
- product. The problem is that I see this as a *very* expensive development
- for two reasons: 1) tcl isn't user friendly (large markets need end-user
- support -- comparing tcl to hypercard, it's hard to imagine how end-users
- would go for tcl) and 2) it would probably take 5-10 man-years (hence $5M)
- and 2-3 years elapsed time to develop TK* and the interactive interface
- builder/development system required to be competitive. That's a big nut
- to crack and high risk to fund.
-
- Nevertheless, I think we should find a way to get one or more companies
- going in something approaching that direction. We already have several
- public domain interface builders. They are fine as far as they go, but they
- need to do much more. Assuming they are ok for the moment, and ignoring
- the programmer-user nature of tcl, the problem to be solved is developing
- TK*. (Incidently, I think the xlib on Mac toolbox and Windows is a good
- idea. In fact, does anyone know if Microsoft has already done the xlib on
- Windows abstraction? I think I heard a rumor to that effect.)
-
- So, what would it take to develop TK*? Suppose we had one library against
- which to code Mac toolbox and Windows versions. The Software Transforms
- folks in Silicon Valley claim to have such a library. I have several
- questions:
-
- 1) How many man-years would it take to port to that or an equivalent interface?
- <My estimate is 3-4 man-years, which includes coding, testing,
- documentation, and support.>
-
- 2) Who would be willing to fund this development. Suppose the cost was
- $300K to get going. If we could put together a consortium with 6 companies
- each paying $50K, we might be able to do the job.
-
- 3) How much would you be willing to pay for the run-time library?
- <My belief is that we must keep the cost low to encourage everyone to
- use it. My thought would be something on the order of $500 for a
- developer license and no cost for distributing applications that use it.>
-
- As many of you know, John and I have started a company that is producing
- low-cost end-user applications using tcl/tk. I am confident that he and
- I could hire the necessary people and do the product development required.
- Problem is how to fund it.
-
- Anyone wanting to discuss these ideas at more length is invited to email
- me directly (larry@cs.berkeley.edu) or call me (510-642-5117).
- Larry
- p.s. If you haven't taken a look at how tcl/tk can be used effectively in
- an end-user application, take a look at our presentation package Perspecta
- Presents! You can FTP a demo version from ftp.uu.net in directory
- vendor/perspecta. In particular, you can look in the lib directory included
- in the distribution for a variety of routines written in tcl. The current
- versio is 42K lines of code of which 11K are in tcl.
-