Programmer 7500

home *** CD-ROM | disk | FTP | other *** search

/ Programmer 7500 / MAX_PROGRAMMERS.iso / CLIPPER / MISC / FPC225_1.ZIP / NOTES < prev next >

Wrap

Text File | 1988-11-10 | 58.7 KB | 1,321 lines

Notes on F-PC Part #1 1. Modified ?DEF.EXT to only append an extension from the word DEFEXT if there is a string in DEFEXT, else doesn't append an extension. 4/6/88 10:47 AM 2. Noticed a bug in SED, prints DONE when finished with a PASTE, on the second line, shouldn't do that. 4/6/88 10:47 Fixxed in PRINTING.SEQ 4/6/88 11:19 AM 3. Removed the public domain notice from the SED exit message. This is not to imply that SED is not public domain, it was simply a distraction, and was not needed. 4/6/88 11:41 4. Added paste date and time with "Alt-O P". The paste is always done in insert mode, to avoid accidental overwrite of existing text. 04/06/88 13:27 5. Re-arranged the SED prompt a bit, to make it clearer. also noted a bug in the WFL window pop up, it was leaving an extra item on the stack when you enter it. Found an old leftover from some earlier code still around, got rid of the culprit. 04/06/88 17:23 tjz 6. Modified FSAVE, and added TURNKEY to F-PC. TURNKEY will now automatically save a system without heads, and with a minimized LIST space. Use TURNKEY as you would FSAVE, followed by a name, but first setup BOOT or DEFAULT to perform your application, and then leave back to DOS. 7. F-PC is now setup to allow you to reduce the #CODESEGS to less than 4096 (=64k), and save some of the code space that is not needed. The editor "Z" for example created from XF only requires about 220k to run, down from 400k for XF the development system. Yes I know its still large, but that is without removing any of the utilities in the system, much more could be saved by stripping down the system 8. Moved inline " (quoted) strings back into the CODE space, for program compatibility. Have added "" (double quote) which will put an inline string into COLON space. Don't worry, the string is moved to CODE space before its address is returned to you. ------------------------- Version 1.01b ----------------------------------- 06/01/88 13:26 9. Changed the name from F88 to F-PC in all files. 10. Eliminated XTYPE from the editor, since it was a hold over from the time when (EMIT) did not use handle write. Since it was redundant with (TYPE), it is no longer needed. Also removed associated word NORMTYPE, with is also not needed. 11. Fixed the bug in the editor cursor positioning, which would place the cursor in the wrong place if the text extended beyond 114 characters. 12. Fixed bug in MACROS, I was allocating 16 times too much memory for the macro buffer. I forgot to put in a U16/ before the ALLOC command. 13. Fixed bug in ;USES, which was preventing it from working except in the meta compiler. 06/01/88 15:42 06/02/88 17:57 14. Working on VTYPE, have changed the name to VIDEO-TYPE. It now needs only ADDR and CNT, X and Y are no longer needed. 15. The editor now displays 80 columns across, rather than 78. I have removed the border characters on the left and right sides. This was done primarily for performance reasons, to make F-PC FAST on a PC speed (4.7mhz) computer. 16. Added ON> and OFF> to the EQUCOLON file, for turning on and off any variable or constant. They save one pass through NEXT, and so are faster than VARNAME OFF or VARNAME ON. I know it seems silly to save such a small amount of time, but I am finding it useful. 06/02/88 18:24 06/03/88 12:46 17. VIDEO-TYPE is finally fast enough to redisplay at 10 screens per second on an XT speed (4.7mhz) machine. I'm going to stop working on it now. I have blown out the look in VIDEO-TYPE for that last ounce of speed. 18. Restructured PASTE (Alt-V) in the editor, so it is much faster, I got some complaints about it's original speed, it now pastes lines from the paste file at 500 bytes/sec on an XT, and 3000 bytes/sec on an AT. 19. Fixed minor bug in NEWFILE that was leaving 2 items on the stack when the file was actually created. 06/03/88 16:57 Assembler bug fixxes 20. Fixed bug in TEST instruction, for immediate mode, changed instruction from A4 to A8. 21. Fixed bug, conditionals were not forcing a compile of the previous instruction. They do now, and you can put conditionals on the same line as normal assembler instructions without any problem. 22. Corrected a bug in the initialization of the APRIOR array, CODE was not clearing out APRIOR properly. This was preventing the retrying of assembly code after an error was encountered. Once an error was detected, then the bug would cause an error EVERY TIME. 06/03/88 17:01 Editor mod. 23. Adjusted the fill line character for lines displayed after the last lines of a file from a blank to a B1 fill character, which is a 50% gray. It is now easy to see which is the last line of the file. 06/03/88 17:13 DUMP mods 24. Added the absolute segment to the beginning of each dump line. 25. Modified XDUMP, so that is takes a SEG value which is added to the XSEG, and dumps OFFSET relative 0 from that point. This makes it actually useful for looking at the body of a definition as follows: ' HEX >BODY @ 10 XDUMP <enter> This will dump the first 10 butes of the BODY of the definition HEX. LDUMP can then be used to dump at other relative offsets from there. 06/06/88 09:40 26. Added out of branch range error checking to assembler in the label mechanism. 27. YDP and XDP have been moved out of the USER VARIABLE list. This system does not support multiple users anyway, and placing these in the user variable list causes problems in other areas. 28. Add L$ and L$: to PASM.SEQ, to allow forward branch resolution of JMP and CALL instructions with distances greater than 127 bytes. See PASM for more information. 29. Finally found and corrected the BUG in EXEC.SEQ, that was preventing SYS from working on a DOS 2.x machine. That is fixed and DOS 2.x can now spawn a shell properly. 06/06/88 18:05 06/09/88 10:57 30. Added checking on COPY (Alt-C), CUT (Alt-X) and WRITE (Alt-W) to see if the file to be written already exists. If the file to be written does exist, then the user is prompted for whether to overwrite the file or abort the write. 31. Corrected a small bug that would cause an edit of a new file to start on line two rather than on line one. This was minor, but still irritating. 32. Added RECOVERSCR which is like RESTSCR but only gets a copy of the most recently saved screen, and can be called over and over again to restore the same screen. 06/10/88 16:23 06/13/88 10:15 33. Commented out FIX, as it has a name conflict with FIX in the floating point package. You may un-comment it out if you wish, it is located in the file TOPEDIT.SEQ. 34. Fixed a bug in the assembler in the instruction JC. Changed its constant from 76 to 72. 06/14/88 12:21 35. Fixed >NAME. It is supposed to return the name field of NO-NAME when it fails to find the name field of the address specified. It was returning a garbage address. This has been corrected. 36. Added INLINE and END-INLINE to the assembler. These give you the ability to add in-line assembly code to a : (colon) definition. See the ASSEM.TXT file or PASM.SEQ for an example of usage. 37. Corrected a feature of FPRINT, FLOOK and EDITALL which prevented them from working on any drive or directory specified. They will now print, look or edit any matching files in any specified drive and directory. 06/14/88 15:54 06/15/88 10:41 38. Fixed the definition of CREATE in the USER vocabulary, it is supposed to create a header, like the normal CREATE, but was defined improperly. 39. Extended the user variable area by 20 bytes, which is room for 10 new 16bit user variables. Previously there was NO ROOM for ANY additional user variables, which seems somewhat limiting. 06/16/88 09:29 40. Fixed a bug in the editor that was allowing you to try and edit a .BAK file. This needs to be prevented, since the editor uses .BAK as the extension it uses for backup files. ------------------------- Version 1.02b -------------------------------- 06/16/88 10:03 41. Changed F-PC to version 1.02b (Beta). 06/17/88 18:58 42. Re-coded GET_ALINE in assembly, as a step to make the compiler faster. I know its silly, but I just can't give up. 06/20/88 16:19 43. Fixed a bug in ACTIVATE. It was missing a DUP. Thank you Dr. Ting. 44. Fixed a bug in UNDEFER. It simply didn't work. Thank You R. Smith. Removed UDEFERS, and consolidated its function into DEFERS. Expanded UDEFER to deal with USER DEFERed words. Moved DEFERS and UDEFER out of KERNEL3.SEQ to DEFERS.SEQ. 06/21/88 18:03 45. Thanks to Roger Bird for an improved SEARCH routine, it is 20% faster, in the normal case, and 10 times faster in the worst case (leading space in search string). 06/23/88 10:35 46. Found and fixed another bug. If SETVIEW is set to view the source from another directory, and you have loaded (FLOAD or OK) another file, you will not be able to VIEW the new files words, because SETVIEW is set to another directory not hte current directory. The solution was to add s constant called WITHPATH, that determines whether the files PATH will be included int the file VARIABLE that gets created for each file loaded. I set WITHPATH to false while loading the Forth system, then set it to TRUE just before saving the system, so all user files will include their PATH for viewing. WITHPATH controls whether "LOADED," includes of skips the files path. The user need not worry about any of this. 06/24/88 11:22 47. NOTE ******** Removed Graphics.SEQ from SMITH.ARC. Also updated SMITH.ARC with Robert's latest version of software floating point. Added Smiley's archive to my high density archive disk. This will be given to the FIG librarian, but is too large to be included with the F-PC system unless an additional disk is added. 48. NOTE ******** A preliminary version of the F-PC Users Manual published bye Ting is available. Contact Dr. Ting for information on availability and cost. Offete Enterprises, Inc 1306 South B Street San Mateo, Ca. 94402 Ph# (415) 574-8250 49. Fixed a bug in the decompiler, that did not clear the lower lines of the screen between the decompile, and the scroll section. The bug associated with decompiling definitions that are too many lines long when in the debugger, is still there, I have not found a good solution for this as yet. Also the X command in the debugger is not yet a toggle, but may be one of these days if I can get around to making it so. 50. Modified the DIR command, to allow multiple arguments on the line after DIR, like "DIR /w *.txt <enter>" will now work properly. ------------------------- Version 1.03b ------------------------------ 06/27/88 17:12 51. Re-coded ABS, DABS, & DNEGATE to improve their performance. This was done on the recommendation of Robert Smith. 52. Added the definition K which is the third outer loop index. 53. Removed the constants 0, 1, 2, & 3 from the dictionary since constants are slower (29 cycles for constants vs. 25 cycles for literals). This also solves some problems with some programs that accept input from the operator, and need to differentiate numbers from dictionary entries. 54. Re-coded <> in assembly for performance. 06/28/88 13:32 55. Bob found a bug in SEARCH that caused it to be eratic on an XT type machine. Fixed by removing the instruction sequence "POP BX SUB SP, # 2" from the code, and replacing it with another technique. Also modified the SEARCH setup to remove a sensitivity to whether CAPS contained something besides -1. 06/29/88 15:55 56. Found a bug in Paste Text, that caused the pasted text to overwrite the line into which the text was pasted, which resulted in ONE lost line. Made a siple fix to PRINTING.SEQ in the word IMPORT. 57. Added a symbol back into the editor to tell you where a page break will occur. The symbol is the same down pointing triangle, but is located on the right margin in reverse video. 06/30/88 11:06 58. Finally found the real cause of the bug that was causing files to be deleted when thay wern't changed when using EDITALL. It seems the length of the file handle was not getting setproperly, and when I changed the file extension to .BAK, it was putting the .BAK extension way out past the real filename. IT IT REALLY FIXXED NOW!. 59. Did some minor optimization, changing double INC's and double DEC's to ADD or SUB instructions to save a few cycles. 07/02/88 00:12 60. I did some benchmarks recently, and found that F-PC is 2 1/2 times as fast as my previous Forth named PF, and over twice as fast as the original F83 system in compiling. 07/05/88 15:42 61. Saved 17 cycles in the combination of NEST and UNNEST. Also saved a some cycles in other words, like OVER, 2OVER, R>, 2R> ect. 07/07/88 09:54 62. ******* Exciting News ******* I have found a way to build the system with INLINE NEXT, and still have the debugger work. This results in almost a 20% improvement in overall performance. I have seen compile speeds over 25000 line per minute on my 16mhz 80386, and over 12000 lpm on a 6mhz AT. See the file DBGFIX.SEQ for information on how this is accomplished. 63. At Ting's request I have added the word INCLUDE as an alias to FLOAD. I have also dd the comment pair /* and */ which can enclose comments. Some people think they are better. 07/07/88 17:37:04.05 64. At the request of Jay McKnight I have generalized the date and time formatting. The word FORM-DATE takes a double IBM format date, and formats it into ascii in the buffer DTBUF, according to the defered function in BUILD-DATE. FORM-DATE returns a counted address which can be typed or printed. There are now three date formats supported, M/D/Y Y-M-D and D.M.Y. Entering any of these makes BUILD-DATE construct that date format. FORM-DATE, and FORM-TIME both build their date or time in the DTBUF array which is 16 characters long, so they must be used then typed or printed sequentially, or they will overwrite each other. ------------------------- Version 1.04b ------------------------------ 07/08/88 14:13:35.77 65. Changed version to 1.04b. 07/08/88 17:12:15.74 66. Per a suggestion from Jay McKnight, I have modified the file printing, such that the actual file date is printed at the bottom of each page along with the current print date. 67. Coded CNSRCH in assembly to make the decompiler go a bit faster. 07/09/88 18:49:17.18 68. Added word wrap, and paragraph justify back into SED. All I had to do was extract the code from ZED, and paste it back into SED. Unfortunately there is still some work to be done before they will be useful, there is currently no way to set the right margin. It just defaults to 132 columns. Justify currently tests for a right margin greater than 130 characters, and refuses to do anything until you have set the right margin. I will try to get to this shortly. 69. Split SEDITOR.SEQ into two files, the end portion being named SEDIT2.SEQ, and containing the upper portion starting at the function table and continueing to the end. 70. The upper limit of file size has been adjusted slightly to 64500 characters. 07/10/88 21:42:07.28 71. Added a method to set the right margin, use Alt-S then R. The Alt-S is for SETUP, and R is for right margin. "R" is the only thing in setup for now, but more may follow later. Control-B paragraph reformat is back in from ZED. Word wrap and Paragraph reformat have both been placed in a file called SEDITWP.SEQ for word processing extensions to SED. If this file is not loaded, then the commands for word wrap, paragraph reformat, and right margin wont work, but won't crash either. 72. This is probably a good place to mention that the editor is split into multiple files of which some do not need to be loaded. For example, the file PRINTING.SEQ contains the print routines, as well as the cut and paste functions, which don't have to be loaded if you are short on space. The EDITHELP file can also be omitted as can SEDITWP if you are short on space. These won't save you a lot of room, but at times every bit can help. 73. As you use the editor you will probably notice that there are several different user interface techniques used to get information from YOU the user. This has happened primarily because I am using SED as a vehicle to test ideas. Some examples of this phenomenon are the print parameter screen, the popup file selection window, and the top line prompting for things like Alt-O, and Alt-S. I realize these are significant inconsistancies, but I haven't had time to clean things up. Perhaps one day I will, lets hope! 07/11/88 11:02:51.38 74. Added UNDO per Robert Smith and Mark Smiley. UNDO was approved by the ANSI Standards commitee at their May 1988 meeting. 07/11/88 19:20:20.14 75. Installed a new help facility in the editor, you can now choose to either have the editor help compiled into the dictionary, at a cost of about 5k of list space. Or you can have the help text brought in from disk on request, if it is available on disk in the VIEWPATH at the time it is needed. Or you can choose not to load the help file at all. There is a boolean flag near the beginning of the SEDHELP.SEQ file that can be set to FALSE if you want help to be in a disk file that will get read in as needed. 07/12/88 15:57:12.62 76. I have added some new files to the F-PC archive, these files contain some functions from the old ZED editor. The first file is SEDSORT.SEQ which adds paragraph sort to the SED editor. It is assigned function key F7, and is mentioned in SED help screen 9. The other file is SEDDRAW.SEQ which add ZED's character line drawing ability to SED. It is assigned function key F9, and is also mentioned in help screen 9. Also added SEDAPND.SEQ which allows marked text to be APPENDed to TEMP.SEQ or the file you specify. 77. Added a disk read buffer size adjuster to the install list. This allows you to customize the disk read buffer for your speed of computer. ------------------------- Version 1.05b ------------------------------- 07/19/88 10:29:32.04 78. Removed SHNDL which was a variable, and added SEQHANDLE which is a constant. They perform the same function, but since their interface changed, I changed the name to protect the innocent. Instead of storing into SEQHANDLE with "SHNDL !", use "=: SEQHANDLE" and of course SEQHANDLE simply returns the value of the current stack file handle. I have also renamed .SHNDL to .SEQHANDLE, and SHNDL+ to SEQHANDLE+. I am sorry for the inconvenience, but the program usage reads much better. 79. I have made several small changes internal to the editor, in preperation for modifications to allow larger files to be edited. 80. DESIGN NOTE: I am finding it easier to read programs that don't contain @'s and !'s. Much of my inclination in F-PC is to use CONSTANTs rather than VARIABLEs for readability. I then use "=:" to assign into the constants. In fact performance is improved slightly. 07/19/88 12:40:14.02 81. Minor enhancement --- Added NEEDS back into F-PC, I removed it some time ago, and I don't know why. 07/20/88 17:27:50.74 82. MAJOR NOTICE ********* NOTE ********** I have modified 2>R and 2R> to be compatible with the way Forth Inc. does them, that is the order of the values on the parameter stack is maintained when they are moved to the return stack. So 2>R is NOT the same as ">R >R", it is the same as "SWAP >R >R". This may make problems for some people, but it is in some sense more compatible with existing (existing?) standards. 07/22/88 10:08:17.33 82. Minot bug fix --- Fixed bug in assembler generated full word immediate data when it should have generated a byte of data in arithmetic instructions in the 7MI instruction class. Example ADD AH, # $80 would generate the proper instruction, but the data assembled was 16 bits instead of 8 bits as required. This caused the following instructions to be very messed up. 83. NOTICE THESE NOTES --- Robert Smith and myself have just received a disk from Mark Smiley that contains a text file of some comments from the east coast Forth board from people using F-PC, or its cousins. I would like to address some of the things mentioned there, but first here is a note about the suggestions I am including. As suggested I have added the attribute words >ATTRIB5 - >ATTRIB8 to allow a greater number of attributes to be available on machines that support them. >ATTRIB5 thru >ATTRIB8 are the same as >ATTRIB1 to >ATTRIB4 for color. For monochrome, >ATTRIB5 is set to >BOLDBLNK, >ATTRIB6 to >ATTRIB8 are set to be the same as >ATTRIB2 to >ATTRIB4. Several people objected to my use of CONSTANTs as NOT constants, that is I like to change them after creating them. I under stand the objections, and I have accepted the suggestion from Steve Palincsar of the East Coast Forth Board (ECFB) to use the word VALUE for any constant that is likely to be changed later. The system has been updated to use this concept, and the decompiler has been updated to show VALUE for values. Another major objection several people seem to have to F-PC is there is no BE-HEAD mechanism, and I seem to throw all of my junk words into the HIDDEN vocabulary. I plead guilty as charged, although I have no simple solution. The reasons there is no BE-HEAD mechanism in F-PC are as follows: 1. I wanted anyone learning F-PC to have FULL access to all of the words in the system, and BE-HEADING some of the words makes it more difficult to learn how the system works. That is NOT my goal. 2. The mechanism used in F-PC/F83Y to get back to the head of a definition from the Code Field is to hash the CFA into one of 64 threads, and search a limited area of head space for the desired CFA Pointer. If you remove headers from the system these threads become messed up and you cant get back to the head from the CFA any more. Consequently the the decompiler doesn't work if you do this. I have not had a chance to look at the BE-HEAD file I received, but if it solves this problem, then I will consider including it with the system. Finally MANY people object to the fact the system changes so much. To these people I can only say that until recently I had no desire to make the system stable, I was evolving it toward what I saw as my Ideal Forth System. As I make major changes to the architecture I changed the name. Now that F-PC has stablized in the sense its major internal structure is not changing, portability will be enhanced, and you won't have to deal with with such a bowl of jello. However F-PC is still in BETA test, that is is is still chaning. Within a couple of months it will be out of BETA, into a full release, then you can complain about stability, until then, YOU GOT JUST WHAT YOU PAID FOR. This may seem a cynical point of view but of course you are more than welcome to make a system you like and release it. So as not to end on a soure note, I do appreciate all the work people are doing with OUR system, and I think this file proves F-PC is (at least a bit) more stable. Send letters or comments to: Tom Zimmer 292 Falcato Drive Milpitas Ca. 95035 Home 408/263-8859 Work 408/432-4643 Updates to F-PC are available at the above address, include a mailer, a high density disk, $25.00 and THREE STAMPS (first class). 07/22/88 17:49:28.73 84. Minor bug fix --- The words which enter the editor all now force DECIMAL as the current BASE. 85. Minor change --- FIX which was removed a while back is back in as REPAIR. --- FIG July Version ended HERE --- 07/25/88 17:10:14.02 86. Significant enhancement --- Added Printer control with the PRTCTRL.SEQ file. If you look at the file PROPRINT.SEQ, you will see the definition of a printer, that is the default for F-PC. duplicate this file, and change the string value and names to conform to your printer. Having done this, load your new file, to make it take effect, and save F-PC back to disk with the printer installed. The changes allow %B to turn ON and OFF BOLD, so: %B This will be BOLD %B The same can be done with %U UNDERLINE, and %1, %2, and %3 the user definable functions. Function %1 defaults to DOUBLE WIDE printing on the provided PROPRINT.SEQ. The file PRINTING.SEQ is setup to look for an existing definition for the word PTYPE. If it does not find one (as in the case where PRTCTRL.SEQ was not loaded) then it will define PTYPE with a generic (just print the string) routine. This means you don't have to load the file PRTCTRL.SEQ if you want to save some space. PROPRINT.SEQ will also automatically NOT load if PRTCTRL.SEQ is not loaded. 07/26/88 14:47:26.31 87. Information --- Received the information from Mark Smiley a few days ago, There were several complaints on the East Coast Forth Board about F-PC not having a BE-HEAD mechanism. Well NO MORE. There was one included in the files Mark sent me, I think it was written by George Hawkins, although there is no indication in the file. I have not included it directly in the system, you will have to load it if you want it. there is an editorial in the BEHEAD.SEQ file that I added, explaining why. Briefly I want YOU the USER to have access to all of F-PC, and BEHEAD interferes with SEE and VIEW. This is a great utility for a professional programmer who needs to save some space to make room for even more definitions. Since F-PC is limited to 5000 or so definitions in HEAD space, BEHEAD could be used to substantially increase that limit to be effectively unlimited. 88. Minor bug fix --- Made a minor modification to -TRAILING to test for the length=0 case. This was correctly pointed out by Mike Sperl on the East Coast Forth Board. 89. NOTICE THIS --- Additional comments from the East Coast Forth Board (ECFB) imply I may be CHANGING THINGS TOO FAST. Sorry about that guys (and gals) Things really are stabilizing hard as it is to believe. I really do appreciate the comments, and as you can see from the above notes, I have implemented some of your suggestions. Of course I have been thinking of COMPLETELY REDESIGNING F-PC (only Kidding). I am very interested in any and all example or utility code you have written, ╔═══════════════════════════════════════════════════════╗ ║ I would like to caution you in one area though, ║ ║ which is as follows: ║ ║ ───────────────────────────────────────────────────── ║ ║ Please don't make any changes to an internal file ║ ║ unless you have to, here is why. ║ ║ ║ ║ F-PC continues to get bug fixes, and if you modify a ║ ║ file and distibute it to people, and that file gets a ║ ║ bug fix (like the assembler has had several times.) ║ ║ then people won't get the bug fix. A perfect example ║ ║ of this is the version of PASM that was distributed ║ ║ with GOTOs in it, it was a neat idea, but now anyone ║ ║ who uses that assembler won't get any bug fixes in ║ ║ later versions of the assembler. If you can make your ║ ║ utilities as add-ons, doit that way. if I need to ║ ║ make something else a DEFERED word, let me know, I'm ║ ║ not completely heartless. I have and am trying to ║ ║ make the system as flexible as possible. ║ ╚═══════════════════════════════════════════════════════╝ 07/26/88 16:35:28.61 90. Significant enhancement --- Thanks to who ever wrote the wait for retrace code for direct video output, I have included a sightly modified version, and it makes a big difference, on a COLOR CGA display. I didn't fix this before, because I did not have a CGA display. I still don't, but the notes on the ECFB brought the problem and solution to my attention. I am writting this note using a CGA display, and I am glad I don't have one myself, even with blanking the screen is still TOO JUMPY for me. The screen save and restore routines have also been modified to perform their operation during retrace. I also made a small change to "L", it used to do a DARK before listing, which causes a terrible JUMP on a CGA display, so now it just homes the cursor, and displays the screen, with erase to end of line. Much les flashing. ------------------------- Version 1.06b ------------------------------- 07/26/88 16:50:41.75 91. I figured it was time for a new version number. 07/26/88 17:56:06.34 92. Minor enhancement --- Added Goto Page per the suggestion from Mike Sperl of ECFB. It is implemented as Alt-G for GOTO, and leaves you on the first screen line on the first line of the specified page. 07/27/88 21:59:16.97 93. WONDERFUL new feature --- I have modified VIEW, so it displays the HELP for a word that is being viewed on the top 7 lines of the display, and displays the source on the next 12 lines. You get to see how to use a word at the same time as you see its source file. HELPON and HELPOFF toggle this feature. HELPON is the default, and if you type HELPOFF, then VIEW lorks like it did before, that is it displays 17 lines of source starting at the top of the screen. I have also removed the DARK from VIEW and HELP, as they caused a jumpy display on a CGA. The new method is much smoother. 07/27/88 22:58:06.15 94. Minor bug fix --- Found and fixed a bug today in the assembler, in the definition of 12MIF. In POSTFIX mode it was assembling the following instruction: 23 # MYVAR #) MOV as if you had entered MYVAR # 23 #) MOV It did work right in PREFIX mode though. By the way this is an ideal time to point out again that you will notice there is nothing in the above instruction to indicate a BYTE or WORD size operation. In POSTFIX mode, these type of instructions default to WORD size moves, BUT in PREFIX mode they default to BYTE size moves. This may seem strange at first, but for compatibility I had to leave the POSTFIX mode defaulted to WORD, and for closer compatibility with MASM, PREFIX mode MUST default to BYTE. Sorry about the confusion, but life is full of this kind of stuff. 95. Minor bug fix --- I also fixed a bug in CURLEY's disassembler, it did not know the difference between a byte move, and a word move, so I modified it to print MOV, for WORD moves, and MOVB, for BYTE moves. 07/28/88 17:46:30.34 96. Minor bug fix --- The META Compiler was causing undefined words to be printed incorrectly at times. Also added some checking to META86 to warn you audibly if there are any undefined references. 07/29/88 16:35:17.90 97. Minor bug fix --- Fixed bug in disassembler that used a literal value to decide when a word was a code word. Modified the decompiler to use a constant MAX-CLASSES that the disassembler can use to test for this case. 08/01/88 15:48:29.62 98. Minor enhancement --- Adjusted SED so it can work within a window. The function Alt-S-W allows you to Setup the Window height of SED for your preference. Using less lines allows you more lines of Forth commands after leaving SED. I know it is hard to understand what I'm talking about, just start SED to see how its user interface has changed. When you leave SED, the last area of the file you were editing stays on the screen. 08/01/88 21:53:54.28 99. Minor enhancement --- I have setup SED so the last edit screen says on the screen after you leave SED until you type DONE, to tell it to start scrolling the screen again. The area that scrolls can be controlled as described above with Alt-S-W (Setup Window). 100. Minor enhancement --- Alt-S-W (Setup Window) now allows you to expand the edit window up to the point where you are using the full screen except for the last and first lines for you edit. The algorithm is difficult to explain, but works well, try it. 08/05/88 15:20:30.87 101. Minor change --- Removed BEHEAD.SEQ from the F-PC archive. I have placed it in the file MISC.ARC It was taking up too much room to allow me to hold all of the source for F-PC and the F-PC.EXE file on one 730k, 3 1/5 inch floppy. 102. Significant enhancement --- Added a variable called BLANKING, which is set during installation to control the blanking of the screen when a CGA or color monitor is used. Some color monitors don't need screen blanking to avoid the snow or sparkle of a normal CGA, so you can specify what you want now in the installation process. 08/06/88 18:35:45.93 103. Minor cleanup --- Fixed bug in Print Window, It was not restoring the screen to the same state it found it, so thing were not as pretty as they should have been. Also made Print window smaller, there was no need to cover the entire screen with it. ------------------------- Version 1.07b ------------------------------- 08/08/88 15:02:29.83 104. Significant enhancement --- Added a menu driver to SED, in the file SEDMENU.SEQ. This initial version has only a few items, but I will shortly be expanding it to include a number of the existing commands. 08/12/88 17:35:35.24 105. Significant enhancement --- Generalized the menu driver, so it can be used at the Forth level. To see an example of yow to create a set of menus, look at the file SEDMENU.SEQ. This file contains the menus for the editor. You can create your own menus, or expand the default menubar by changing MENUS.SEQ. the current default menubar is available in Forth, by pressing ESC, after which you can walk around on the menu bar, and execute a function by pressing <enter>. Pressing ESC again clears the menubar from the screen and restores your cursor. 08/15/88 10:59:24.59 106. Minor cleanup --- Mod the the debugger, allows debugging to continue on with the definition above one you have nested into when you leave the lower definition with an EXIT. Previously the system just continued execution, and did not allow debugging further. This was fixed by Robert L. Smith. 08/15/88 14:21:29.17 107. Minor cleanup --- Fixed X in the debugger to be a toggle as suggested by several people. Pressing it once turns OFF the source display, pressing it again turns the source display back on. 08/15/88 16:31:54.46 108. Significant BUG --- Fixed bug in 2+!, I was adding in the double stack value to memory in the wrong order. Thanks Roger. ------------------------- Version 2.01 ------------------------------- 08/16/88 15:14:01.84 109. Version 2.01 08/17/88 15:46:09.94 110. Nice enhancement --- Add a selection tool to SED for graphic and special characters. Note that only the special characters above 127 can be printed on an IBM compatible printer. Those characters below 32 are useful only for screen display purposes. 111. Prepared the screen dumps for version 2 of the users manual for Dr. Ting. This users manual is being done on a Macintosh and will be laser printed, as such the production quality will be MUCH higher than the earlier manual. It looks like we will have some good documentation for F-PC. I highly recommend buying the users manual, I think you will find it very useful. 08/18/88 16:08:37.92 112. Minor cleanup --- I have made a minor change through out the fabric of the menu system that allows editor functions other than menu names to be performed while the menu is on the screen. This was done to correct a problem that two people ran into, which was you could not do any thing except perform a menu operation when the menu was on the screen. They found this very confusing, and I have thus corrected it. 113. NOTICE THIS --- I have run into an interesting problem, is seems Dr. Ting was working on a program that needed some text to process while verifying a program section. He decided to use the editor text buffer as a convenient place to keep this temp text. To make a long story short, I made a change to the editor which made his test program non-functional. I want everyone to know that the editor is my toy, and it is fair game for changes at any time, and so you should be VERY CAREFUL if you decide to use pieces of it, as even its internal architecture may change without even a moments notice. This file if for documenting changes primarily to the Forth system, NOT the EDITOR except from a user interface standpoint. 08/19/88 00:06:08.27 114. Minor cleanup --- I have added a window entry mechanism to the following routines: SAVEAS, SEARCH, REPLACE, SHIFT-CUT, SHIFT-COPY, SHIFT-APPEND. They now use LINEEDIT, to get their parameters, which allow you to use the keypad to edit/modify your selection. The previously entered value in SEARCH and REPLACE will be displayed on entry. 08/19/88 17:05:17.70 115. Significant enhancement --- Line edit now has available through the use of Alt-A, access to the special character entry box used by the editor. When you are entering a search or replace string, you can press Alt-A, and you will be given the opportunity to enter a graphics or control character. If you use this feature in an application, you will need to add information to your on screen help to let the user know about Alt-A. See the editor search (F6 and F8) for awwn example of how to let the user know about this. ------------------------- Version 2.02 ------------------------------- 08/20/88 15:21:09.32 116. Version 2.02 117. Minor cleanup --- Changed default file specification for the popup file selection tool to "*.*" from "*.seq". This makes it easier to see that you can walk around through the directory tree of a disk. You can of course use "\" while in the tool to change the file spec. 118. Significant Milestone --- I proofed the final master (LASERWRITER printed) copy of Dr. Tings users manual for F-PC yesterday. It is a HUGE improvement over earlier manuals, and should do much to help make F-PC acceptable to a larger audience (I hope). Also much work has been done on the technical reference manual, and I hope Dr. Ting wil be able to complete it soon. It is going to be at least on par with the INSIDE F83 manual he produced several years ago. I am also pleased to report "Ting's English" as it has come to be known, has moved MUCH closer to American English in the last couple of years. The manuals readability is very good. Jay McKnight will is also helping proof read the F-PC Technical Reference Manual. 08/22/88 14:37:58.76 119. I have added a utility to ZIMMER.ARC that allows editing Forth command lines like DOSEDIT, but under control of Forth. It is called XEXPECT.SEQ. All you have to do is load it, and you can use UP and DOWN arrows to recall the most recent four typed in lines. -------------------- Version 2.02 08/24/88 09:20:05.33 ----------------- --- FIG August Version ended HERE --- 08/24/88 10:34:59.12 120. Minor enhancement --- Added a flag to Line edit to determine if trailing blanks are to be stripped from the edit string. The word "STRIPPING_BL'S" gets automatically set to TRUE after being set to FALSE, so you must set it to FALSE every time you call LINEEDIT ns you want trailing blanks "LEFT ALONE". 08/29/88 09:44:25.14 121. Minor cleanup --- Added CLS as an alias to DARK. 08/29/88 15:53:40.88 122. Minor cleanup --- Made a minute change to "ME@", it used to be able to call abort if it could not find the end of the environment, and would on a DOS 2.x system print garbage. It has been changed so it will not abort if it cannot find the end of the environment, it will instead return a null filled ME$. It will also return a null filled ME$ if it is performed on a DOS 2.x system, so it won't print garbage anymore. 08/29/88 22:20:03.45 123. Minor bug fix --- Corrected a small bug in the decompiler that was causing SEE to show the wrong function when SEEing DARK or AT. These are self defining words that it did not handle properly. This has been corrected. 08/30/88 21:57:03.55 124. Bug fix --- The definition of IBM-DARK was always trying to set the video mode to 2, which is of course wrong for monocrome, and was messing up some systems with two monitors. It now reads the video mode and restores the same mode. 09/01/88 10:35:57.67 125. Cleanup --- Changed the error message in ?MISSING from "Huh, I'm confused?" to "What?". I received some complaints that the Huh message was irritating to some people. Also modified ?MISSING to print a space before printing the confusing word to better delimit it from anything already printed on the line. ----------------- Version 2.021 09/01/88 14:06:22.84 ----------------- 09/02/88 21:28:49.05 126. Bug Fix --- Added SWAP to PRINTING line 251, in the definition of GETINPFILE. It seems you could not hold down shift and get the specified file to be inserted, two values were in the wrong order. ------------------- Version 2.10 09/02/88 23:29:31.74 ----------------- 09/02/88 23:29:05.71 127. Major Enhancement --- SED has just been modified to allow editing files as large as memory. Now F-PC is a BIG program, and takes a lot of room, but you can still edit about 160k file on a 640k machine, and a 64k file on a 512k machine. This change took me about two days to complete. It was relatively minor, in that the line pointer table went from pointing at lines with an offset, to pointing at lines with a segment. The text line structure then changed to counted strings. Files are now read and written a-line-at-a-time through the normal lineread mechanism and a new linewrite mechanism that is currently part of the editor. I may move the line write out of the editor later, and into the system. It seems to work fairly well: there is no noticable speed difference. I can make no guarantees as yet about the reliability. I simply haven't had time to wring it out yet. 128. Minor enhancement --- I have added a word UNEDIT to the system. It DEALLOCATES the editors memory after an edit session, so you can use that memory for your program if you need it. The editor automatically grabs all of the available memory except for the amount you tell it not to during installation (32k-256k), so it is useful to be able to recover that memory if you are using an application that uses memory outside of F-PC's CODE space. 09/04/88 17:59:25.56 129. Significant Improvment --- Recoded the way a goto line is performed. I used to goto a line by a line at a time, now I block move the lines in big chuncks, and correct the line pointer table. This is MANY times faster, and now you can move from the beginning of even a VERY LARGE file to the end or visa versa in WELL UNDER A SECOND. This speeds up both starting and finishing an edit of a file significantly. 09/05/88 00:37:28.75 130. Space savings --- All of the installation messages, SED help messages, and the INFO message have been moved into a file called F-PC.MSG These messages are retrieved from the INFO file as needed when the specified operation is performed. This resulted in a savings of over 11k in the .EXE file. If you don't need help in the editor, or can live with short messages during installation then you can delete the F-PC.MSG file altogether. As this was mostly wasted space for normal operation, this seemed like a good feature. 131. New installation setting --- LIST segment allocation has now been added to installation. This makes it easy to readjust your memory usage. Just re-install the program. 132. Editor adjustment --- The editor has been modified to allow an edit to take place even if there is only room to edit a 16k file. This allows the editor to be used in very memory limited systems like 384k (maybe?). If you try to edit a file that is too big for memory, SED will perform a partial read, and inform you it has done so. You should of course NOT CHANGE the file and DON'T SAVE CHANGES if you do. 09/05/88 13:02:20.20 133. Bug Fix --- There has been a bug in F-PC that prevented the compiler from compiling the last line of any file it was loading. This was caused by the way I was testing for the end of file condition, and it has now been corrected. ------------------ Version 2.11 09/05/88 13:02:27.23 ----------------- 09/05/88 20:09:39.37 134. SED feature --- Add a menu item to the FILE menu to allow spawning a DOS shell from within the editor. This can be useful if you need to format a disk, copy or delete some files ect. 09/06/88 15:16:25.63 135. Minor enhancement --- Another menu item to the ADVANCED menu, to allow setting the editor Window size. With windows of less than 23 lines, the editor will stay on the screen and only the lower lines will scroll until you type DONE to cause full screen scrolling to be re-started. 09/06/88 17:54:30.17 136. Memory usage optimization --- Adjusted SED such that when you SHELL out to dos, it deallocates all of the memory it does not need for the current file. This makes the memory available for DOS to use. The memory is reallocated on re-entry to SED. Appropriate error messages are issued if there is not enough room for DOS and the file you are editing. I am even handling the case where you loaded a terminate and stay resident program from the shell. I warn you that is a bad thing to do, and tell you to save your changes and reboot your computer because you have fragmented your memory. 09/07/88 14:55:33.93 137. Cleanup of debugger --- Fixed some minor bugs in the debugger, that would cause dirty displays if you were working on a procedure that was too long to decompile on the screen (more that 18 lines). Fixed another bug that would cause the decompiler to pop up a level in the debugger one word before it should. 09/08/88 11:33:22.32 138. Minor enhancement --- Added a flag to LEDIT.SEQ the line editor to allow the programmer to specify whether the first text character typed will automatically clear the edit line. See LEDIT.SEQ for more information on this. 09/09/88 08:42:11.91 139. Minor enhancement --- Added White to black and Black to white inversion to the SETUP menu at the Forth command level. 140. Another enhancement --- Both the compiler and SED have been modified so the editor preserves the cursor position when you leave and re-enter and edit, including the cursor position on the line. The compiler has been adjusted to make the cursor show up exacly at the point in the file where the compile error occured when you enter the editor. ------------------ Version 2.12 09/09/88 15:19:28.70 ----------------- 09/13/88 10:38:12.29 141. Generalized BOX --- BOX and BOX&FILL have been made more generally useful. They now position the cursor on the first blank line of the box just drawn, ready for you to use TYPE or ." to display text in them. You can then use BCR for BOX CR to move the cursor down to the next blank line of a box. See an example in the file BOXTEXT.SEQ. 142. Message cleanup --- Moved most if not all of the SED messages from the status line into boxes. 09/14/88 13:10:55.51 143. Bug fix in Word Wrap --- Stomped out a bug in the word wrap code that was causing extra blank line to be inserted under special conditions. It wan't serious but it was irritating. 144. Menu enhancement --- Well I have made a very nice improvement to the menu structure. Each menu line you create is followed on the same line by the name of the function to be performed. This is much more readable and less error prone than having a seperate array for the functions. Here is an example menu definition: 5 newmenu dfile$ menuline" New file " makefile menuline" Open file " openfile menuline" Edit current file " editafile menuline" ────────────────────" noop menuline" Print Current file " dolisting endmenu Notice how readable it is, and it's very easy to see what functions go with which items. 09/15/88 00:14:03.76 145. Install enhancement --- I have removed the question from the install program that asks you what read buffer size you want to use. The read buffer size is now determined automatically at F-PC boot time by measuring the speed of your computer. It will use a 16k buffer for a 386 class machine, and about a 4k buffer for an XT class machine. It actually saved some space by making it automatic. 146. Minor bug fix --- LOAD did not preserve the edit cursor line. After a LOAD you would be left on the last line of the file if you were to re-edit. This has been corrected. 09/15/88 11:23:33.96 147. Rearrangement --- Moved XEXPECT.SEQ from the optional area of ZIMMER.ARC into the F-PC system. It is now automatically loaded when the system is extended. It is too useful to leave as an optional feature. Use up and down arrow to recover previously entered lines. ------------------ Version 2.13 09/15/88 11:25:18.21 ----------------- 09/16/88 15:01:28.86 148. Auto edit on load error --- There is an optional file now being loaded called EDITERR.SEQ which sets F-PC to automatically call the editor when a compile error occurs while loading. The error message is shown on the lower few lines of the screen and the editor is entered with the cursor placed where the error occured. The editor's display window is automatically adjusted smaller in size to allow the viewing of the error message. 09/19/88 17:02:34.24 149. Enhancement --- I have added a new file and set of functions to the kernel. The file is SAVEREST.SEQ, and the functions are as follows: SAVE> Saves the contents of following work to the return stack. Equivalent to "VARIABLE-NAME @ >R". SAVE!> Save the contents of the following word to the return stack, and set the contents of the following word to the value on the stack. Equivalent to "value VARIABLE-NAME DUP @ >R !". RESTORE> Restore the contents of the following word from the return stack. Equivalent to "R> VARIABLE-NAME !" These words have been inserted wherever they were appropriate in all F-PC files. The .EXE file size actually dropped by a couple hundred bytes, and is I believe more readable. 150. REMOVED SHNDL --- I have removed the error aborting messages for the words SHNDL, .SHNDL and SHNDL+. I switched to SEQHANDLE so long ago that if you haven't updated by now you will have other things to worry about anyway. 151. FUDGE set mod --- Integrated SETFUDGE into BUFSIZE-INIT. The later sets the buffer size according to the system clock speed. FUDGE is now set according to the system clock also, rather than looking at the ROM to try to figure out what computer we are running. Looking at the ROM didn't work most of the time. ------------------ Version 2.14 09/19/88 18:09:59.89 ----------------- 09/20/88 08:37:50.63 152. AUTOEDIT enhancement --- Modified EDITERR.SEQ to add the words AUTOEDITON and AUTOEDITOFF to select whether F-PC will automatically enter the editor when a compile error is encountered while loading. 09/20/88 13:55:20.99 153. Bug fix --- Fixed two bugs. First was in the way the offset into a file was accumulated while compiling. It seems under some conditions the cursor would end up on the wrong line after a compile error. This has been stomped out. The second bug was in NEEDS. It seems it would reload the needed file all the time even after it was loaded. I needed to insert the current drive path into the name being looked for in one of my two tests for having a file already loaded. 09/22/88 08:27:23.88 154. Macros enhancement --- The user interface to Macros has been cleaned up, and the ability to LOAD, SAVE, and VIEW macros has been added. The macros are saved in the file F-PC.MAC. The macro file is not automatically loaded. It must be loaded manually (with Alt-M-L), so you can create temporary macros that won't overwrite your master set. These new functions are all performed from the Alt-M command dialog box so no additional documentation is needed. 09/22/88 13:30:16.69 155. WORDS enhancement --- Added the word THESE, which can prefix WORDS to tell WORDS you want it to only look for matches to the following string in the CONTEXT vocabulary. Here is an example: WORDS X <enter> Looks for words containing X in all vocabularies. EDITOR THESE WORDS X Looks for words containing X but only in EDITOR. 156. Display enhancement --- Added an install section to test the video mode and prompt for whether you are using a real COLOR display of an LCD display. Allows you to setup F-PC for an LCD display that will be usable where the normal COLOR attributes might not be usable. ------------------ Version 2.15 09/22/88 13:33:19.98 ----------------- ******** Further notes may be found in the file NOTES2 *********