THE WELCH COMPANY
440 Davis Court #1602
San Francisco, CA 94111-2496
415 781 5700
rodwelch@pacbell.net
S U M M A R Y
DIARY: December 2, 2008 02:39 PM Tuesday;
Rod Welch
Letter to Steve on assembly for SDS 64-bit program on Microsoft Vista.
1...Summary/Objective
..............
Click here to comment!
CONTACTS
SUBJECTS
Assembler 32-bit Code for Medit Gary Johnson Submits Techno gy Sour
0403 -
0403 - ..
0404 - Summary/Objective
0405 -
040501 - Follow up
040502 -
040503 -
040504 -
040506 - ..
0406 -
0407 -
0408 - Progress
0409 -
040901 - On 040505 Gary submitted a source to research using assembly to create
040902 - the next version of SDS with more memory by updating the current Medit
040903 - program based on assembly. ref SDS 1 QU8K
040905 - ..
040906 - On 081104 Morris discussed writing an updated version of Medit for SDS
040907 - using Java rather than assembly. ref SDS 2 2W6I At that time on
040908 - 081104 reviewed advantages of using assembly, beginning with 30 years
040909 - successful experience using assembly for Knowledge Management
040910 - development, and so requires less time, only about 6 months,
040911 - ref SDS 2 YP4L, and stronger compatibility. ref SDS 2 VO3P
040912 -
040914 - ..
040915 - Sent a letter to Steve
040916 -
040917 - 1. Subject: Editor 16-bit Upgrade 64-bit Microsoft Vista with assembly language
040918 - Date: Tue, 02 Dec 2008 14:45:34 -0800
040925 - ..
040926 - 3. Very kind of you to comment.
040928 - ..
040929 - 4. We know there will be a lot of testing to upgrade Medit one way
040930 - or the other [cited in the record on 081123, ref SDS 3 RO4X,
040932 - ..
040933 - 5. Was hoping to get 2 or 3 advantages for upgrading the editor
040934 - from 16-bit to 32-bit and/or 64-bit assembly, rather than write
040935 - the program de novo with Java, which is the current effort.
040936 -
040937 - [On 081202 2011 Steve responded that recoding Medit will be
040938 - difficult using an entirely different language [Java rather
040939 - than Assembly] environment without needing to then fight
040940 - for perfect compatibility on all of the "edge" cases.
040941 - ref SDS 4 8L8T
040943 - ..
040944 - 6. My colleagues say the only benefits of assembly are small
040945 - program size and fast processing.
040946 -
040947 - [On 081202 2011 Steve cites another significant benefit
040948 - that because Medit is already in assembly language, moving
040949 - from 16-bits to 32-bits is a vastly smaller task that is
040950 - much less prone to translation errors than recoding from
040951 - scratch in an entirely different language. ref SDS 4 8L91
040953 - ..
040954 - [On 081202 2011 Steve recommends recoding in 32-bit MASM so
040955 - that you have universal compatibility both with all current
040956 - PC systems, and with future x64 platforms which will
040957 - certainly be supporting x32 forever. ref SDS 4 YW50
040959 - ..
040960 - 7. They note increasing memory for SDS from 640K to 64 MB makes
040961 - size a non-issue, and running SDS on 64-bit quad-core
040962 - processors will maintain SDS responsiveness using Java, rather
040963 - than assembly. I have doubts. Making the program bigger, and
040964 - increasing the size of work files may reduce responsiveness,
040965 - even for faster processors.
040966 -
040967 - [On 081202 2011 Steve says JAVA is a platform-independent
040968 - solution with JVMs available pretty much everywhere; and
040969 - current machines are SO FAST that JAVA should work well for
040970 - anything that's predominantly interactive and not heavily
040971 - computational. ref SDS 4 8L40
040973 - ..
040974 - 8. My sense is that the current program in 16-bit assembly has
040975 - worked well for 30 years, as SDS evolved through trial and
040976 - error into a robust application.
040977 -
040978 - [On 081202 2011 Steve responds saying keeping the code in
040979 - assembler, just expanding the code from 16-bits to 32-bits
040980 - -- you RETAIN 100% of all of the trial and error work.
040981 - ref SDS 4 8L47
040983 - ..
040984 - 9. Therefore continuing this platform in 32-bit or 64-bit assembly
040985 - introduces the fewest changes that have to be solved for
040986 - getting macros to run in order to maintain continuity of SDS
040987 - development and daily use for Knowledge Management.
040988 -
040989 - [On 081202 2011 Steve seems to concur saying without
040990 - question using assembly provides the best path for
040991 - continuity of SDS capability; and notes that JAVA might be
040992 - more appealing to anyone who is not comfortable with
040993 - assembler, but it's also the case that whereas the straight
040994 - 16-bit to 32-bit translation is LINEAR, the JAVA conversion
040995 - could wind up taking MUCH longer -- it could become a black
040996 - hole as the recoders attempt to get it to work in exactly
040997 - the same way. ref SDS 4 8L54
040999 - ..
041000 - [On 081202 2011 Steve codes in assembler becasue there is
041001 - no compiler producing anything unexpected; enables
041002 - controlling exactly what the outcome will be. ref SDS 4
041003 - 8L6P
041005 - ..
041006 - 10. You seem enthusiastic about assembly, so I thought you might
041007 - have several points that could inform our considerations.
041008 -
041009 - [On 081202 2011 Steve reports writing Windows applications
041010 - in assembler language. ref SDS 4 8L64
041012 - ..
041013 - 11. On your last point about running SDS with Virtual PC, have [you]
041014 - done this, and so know from experience it will work? We have
041015 - been running SDS with NTVD in Windows XP; that is what I would
041016 - first consider using. Any reason to try Windows 3.1 mentioned
041017 - in your letter today?
041018 -
041019 - [On 081202 2011 Seve advises that running SDS in Windows XP
041020 - is fine. ref SDS 4 8M31
041022 - ..
041023 - 12. Sincerely,
041025 - ..
041026 - 13. Rod
041027 -
041028 -
041029 - *****************************
041030 -
041031 -
041032 -
041033 -
041034 -
041035 -
041036 -
041037 -
041038 -
041039 -
0411 -