bson@erice.ai.mit.edu (Jan Brittenson) (03/05/91)
This is the (curtains rise, the audience applauds) announcement of STAR 1.04.3. First, I must thank Gadiel Seroussi and Bob Armstrong for their patient testing of and bug fix suggestions for STAR. Many bugs would never have been found or fixed without their efforts. Gadiel does the testing for MS-DOS as well as provides the STAR.EXE executable for MS-DOS users. Bob makes adjustments and suggests fixes to make STAR comply with VAXC and the VAX/VMS environment. STAR version 1.04.3 can be brought by anonymous FTP from trix.ai.mit.edu as pub/star-1.04.3.tar.Z. Quite a few bugs have been fixed, but a few cosmetic ones remain. Line numbering is now correct in CODE and RPL modes. Local symbols work in CODE blocks. A couple of bugs that caused pointer rot have also been fixed. Here is the updated bug list: o The scapes \ooo (3 octal digits) and \xhh (2 hex digits preceded by an `x') don't work in strings. o Line numbering was fixed in CODE and RPL modes, but broke in lines that cause multiple lines in the listing, e.g. "ASCII `Blaha, blaha, blaha. Talk, talk, talk.'" I just discovered this today and so will let it pass just to get this out the door. As a result, error messages may end up at slightly higher line numbers (a pain in the butt for GNU Emacs users who parse them to jump to the file and line). o Code fields in listings tend to get a little skewed by macros that generate more than one line, but only sometimes. o The current address jumps a little if a macro results in more than one line of code. This is only in the listing, the code and all address calculations are still correct. The former I just discovered; strange how you miss testing some things sometimes - only serves to show how important thorough and automated testing is. The latter was deemed complex and of low priority; it will probably go away with the major recoding and speed-up that will be done in version 2. (There will not be any 1.05.) There is also the slight misfeature that STAR may sometimes lock on MS-DOS. This seems to be a bug; it can be avoided by not refering to symbols outside a CODE or RPL block from within it, should it occur. Included in the tar file is starref.tex, which replaces the old star.ref file. It's LaTeX, and as soon as I have figured out how to convert it to PostScript I will put it up separately for FTP - it will be announced here on comp.sys.handhelds. Included is also dotests, a unix sh script that runs all the test files with some information about how many errors to expect. Gadiel Seroussi modified this to DOTESTS.BAT which together with the SED2Q.EXE program will do the same on MS-DOS. This concludes the development of STAR version 1, except for possible minor bug fixes of 1.04. Version 2 will be significantly different, boasting features such as system-sensitive command line parsing. There will be some linker capability, symbol table files, jump referencing, and support for an ML debugger. It will be given an Finferior mode to let you write ML programs in GNU Emacs and have it automatically syntax checked. The Emacs mode will also support indenting, commenting, and comment filling. Provided will also be a script or program to generate a TAGS file. * * * If you don't have access to FTP or can't send e-mail to Wayne Scott's mail server (wscott@ecn.purdue.edu; Subject: HP-MAIL-SERVER; the command "SEND HELP" in the message body), then you can send me a check or money order for USD 20 (3 1/2") or USD 16 (5 1/4") and I will mail you an MS-DOS formatted diskette with the 1.04.3 distribution as it is on it, including the STAR.EXE executable. If you'd like some particular format (SS, DS, etc) the make sure to say so, otherwise I will rely on the wisdom of a local MS-DOS oracle to choose a default format for respective format. The difference in price is not only the difference in over-the-counter diskette prices at quantities of 10; it also provides me with a way of figuring out what format to ship when none has been specified. If an extra $1 is included, then I will also file your name and address and send you a paper mail announcement and description of features when version 2 is available. The same announcements will be made to comp.sys.handhelds. (Don't hold your breath, though. It may take months, as I have a few other pet projects going on.) Both prices include air-mail postage. People in Europe should get in contact with Wlodek Mier-Jedrzejowicz (umapd@cc.ic.ac.uk) at U.K. PPC - he can supply more details. I will also include a copy of a print of starref.tex - this by itself is not that exciting, your creative talent with a copier and a pair of scissors will have to be put to use. Anything in the distribution can be freely copied or sold, given that you inform the buyer where to get the source distribution for free. If you make changes or include parts of STAR - this applies only to the C source code and hp48.star - in your own software, then the final product, like STAR, must be distributed and available under the GNU General Public License, which is described in the file COPYING. Finally, no manual is yet available. The prefered paper mail order address is (continental North America only, please): Jan Brittenson c/o Liz Highleyman, rm 772 MIT AI Lab Cambridge, MA 01379 USA Don't forget your RETURN ADDRESS. Again, give me time to dig up a PC, to do the copy work, and a few days of procrastination. STAR is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. This applies regardless of whether STAR was ordered by paper mail at a nominal charge, or whether it was acquired by electronic means. Happy hacking, -- Jan Brittenson bson@ai.mit.edu Read my lisp: no new classes!
umapd51@cc.ic.ac.uk (W.A.C. Mier-Jedrzejowicz) (03/14/91)
Jan says people in Europe should contact me for more details on Star ;-) Thanks Jan! What he means is that he has posted me a uuencoded version which I can forward on request - on condition there are not too many - if you have a local user group I'd rather forward just once for local redistribution. My user id here is "umapd51" - Jan's message lost the 51 ! Regards to all, Wlodek Mier-Jedrzejowicz