JII-I Ii Approved For Release 2002101129 27 September 1968 MEMORANDUM FOR: Executive ?Assistant, 0053 SUBJECT I ASCII Standard (NBS Draft) 1. Consider this as an addendum to the "memorandum of 25 September 1968, same topic. 7 2. Another point that hasn't been mentioned in any of the proposal discussions concerns binary data. 3. Most third generation machines have the option of internally addressing bytes or words. In particular, scientific users use both integer (pure binary) and floating point (an encoded binary) arith- metic and these formats use single (four bytes) or double (eight bytes) words. 4. I believe it is important to be aware that these formats are neither EBCDIC and that they do use the eighth bit of the byte. Thus, data in this form is not transmitted over lines in ASCII format without a code/decode process. 5. I overlooked this condition and so might others. It may be helpful if this bit of a reminder is kept in the file so that the possible presence of binary code is considered on future projects. -9 od Release 2002/01/29 . Tip-var: . Approved For Release 2002/01/29 25 September 1968 MEMQRANDUM FOE: Executive Assistant, 063 SUBJECT Recommendation on Draft of NBS Instructions for Application of the Federal Standard ASCII Code 7 REFERENCE Mam. did 13 Sep 63 and Nee Memo 4 dtd4Sep 68 Reoommendation I do not recommend that the Agency endorse the NES draft. ?0 The implementation guidelines cannot and will not be followed by agencies in even a. token manner. The NBS guidelines impart a ?thrust to President Johneon?s memo which is not prudent. Standards 4 '7 previously- produced lay-the U.S.A. Standards Institute have been highly respected and quickly implemented by Government and industry. These standards have been aids to less confusion and more efficiency and more economy. A premature and pressured implementation of this standard on Government users hy NBS will produce confusion, inefficiency. and more cost. The respect which agencies have for NBS and USASI will deteriorate and this will be unfortunate in future "pressured implementations" of standards. I recognize that the guidelines. have infinite loopholes and waivers. but these create ambiguity as to?the force of the guidelines and this is bad. Why bother? All instructions pertaining to standards should be as clear as possible. - At first impression. it may appear that the above position is inconsistent with previous recommendations. It is not. -However, the picture is exesperatingly confusing.- and thus the following infor-_ mation and discussion are. given. - - Pertinent InformetiOn Items 7 1. An increasing majority of computers and data bases are in Approved For Release 2002/01/29: CIA- RDP78- 04723A000100150038-9 era-N4 the" . 7 y?uni EBCDIC, an eight bit code which was created by IBM for third generation equipment. All (or nearly ail) eight hit byte machines of other manufacturers use this code. Because no STANDARD eight bit code has been specified and approved by USASI (United States of America Standards Institute). EECDIC has become the .de facto standard of the computer industry. 3 .4 - - Approved For Release 2002/01/29 CIA-RDP78-04723A000100150038-9 4 .7 Z. ASCII (USA Standard Code for Information Interchange), 1: the USASI approved code for information interchange. ie a seven bit code. Note that this code is designated by definition for information interchange, not for using in computers or for data storage. Granted, i I it would be nice if this code were 100% compatible with that used in computers or in data eterage.? This is not possible. Most new cemn puters have eight bit bytes {not a seven hit byte code) and the re- Vmaincler have six. bit bytes. New style tapes have eight bit byte for- 'mat (nine channel with one being a parity bit) and old tapes have six bitbytea (seven channel with one being a parity bit). Nearly all other magnetic storage for data files. disks, drums. data cells, atrip files. etc. have eight bit bytes and have not been considered in the standard implementation. I . 3. USASI has approved only one transliteration of seven bit bytes . to eight bit bytes and that was recently in X3. . . - NRZIL Note that this standard states how to code ASCII on one kind I (800 of magnetic tape only. This standard is feasible. By making a computer pace through a utility program. tapes written in EBCDIC can be converted to this code, and presumably the recipient of the tape can then reconvert it to EBCDIC no hecan Why do the conversion and reconvereion? This can be rationalized only in that ?it is a standard, and. in that it has never previously been imposed an a mandatory replacement for EBCDIC. 4. USASI has not approved a transliteration algorithm for placing seven bit byte ASCII on an eight bit byte computer. How can the guide! 7 line to? use ASCII in the computer? be implemented by either manu- . factor-er or customer without specs? - 5.. No transliteration of seven bit ASCII to eight bit byte has been approved for other storage devices. am ignoring 1: 1e transliteration problem on non- 300 C. P. I. tape since standards for these probably will be approved shortly.) . i ll .. . ?more. Approvedfpi Release 2902/01/29 CIATRQEW-Q Lilli"- lL 14? 5 [lull Approved For Release 2002/01/29 CIA-RDP78-04723A000100150038-9 6. Collating soqucncos between AEECII o. if the: 800 (LP. 1. ASCII is assumed for all eight bit byte data bases) and EBCDIC are different. Most data bases go through a sorting process at times. Every aoplication would have to be scrutinized in detail and very possibly. most applications would require conversion programming. Is anyone so bravo as to suggest another programming conversion effort? 7. In 1964, the was offered with the internal option of I 13130ch or ASCII-3 own transliteration version of seven bit bytes to eight bit bytes}. Coincidentally. all early customers opted for EBCDIC. chiefly becauoo it woo the more compatible with BUD. the code usod on the older Inachinos. Subchuontly. 1966?1968, IBM quietly ceased to offer the: ASCII-8 option, and used the subtle per- ouodor of "incompatibility to other to ASCII-8 rcquootcrs. The fact that USASI had never approved a. seven to eight bit trans- literation was a. clincher. and no IBM 360's were ever delivered to customers which internally use ASCII code. Also, it is doubtful if third generation machines of any manufacturers have been so endowed. Thus. tho requirement of ASCII internally would introduce new hard- ware problems and more Shakedown of third genoration hardware. 8. After manufacturers failed to produce ASCII compute-re. the oyotomo software: designers gradually forgot about the ASCII option. It is a non): certainty that no present systems software {and much others) would run on an ASCII machine. Would any user be willing to implement a. new OS (Oporating Systoml?Bo? in ASCII and presumably 7 repeat the implementation agony? 9. A final but ironic point. USAS-I is puohing procedural lan- guagoa and no one argues seriously against tho concept. COBOL and FORTRAN are being standardinod. However, two facts are fairly certain, 1) Neither COBOL nor will be. the future general purpose procedural language. and 2) this future language will be a language very similar to or to a annex-sot of and in fact possibly will be PLs-l is powerful and can manipulate bits nearly as well as an assembly language. Thug, the language power can bring no full circle back to language capability and . also to its problems. Tho capability to manipulate bits may mako a. program written to process data bases incompatible for use: if the data ?base were changed to ASCII. (This problem also exists in ALC but to a much lesser extent in and FORTRAN.) r; 3 Approved . ma 7.7 at? me'?mm any? .STATINTL Approved For Release 2002/01/29 CIAFRDP78-04723A000100150038-9 i This locompatibility is superimposed on the collating sequence difference and thus potentially becomes a serious problem. Discussion No agency with much thiro g?noration hardware can follow the guidelines. The specs and the guidelines are too late. A standard is essential when a file is described. It is particularly essential when communicating with outsiders. It ia not important that both partioa use it is important t. at they communicate clearly and precisely. This can be accomplished by specifying or 7 If a receiving installation uses a. code different from tho sending, he can convert. Utilities are available for this purpose. slVory likely. the latter condition la academic: since most lnotallationa are and will be locked to EBCDIC. None of the above negates the importance of ASCII for information 7 interchange. Moat manufacturers have circuitry intho communication line control boxes which automatically transliterate eight 'oit bytes in the computer to seven bit bytes for the transmission and vice versa on return. One of the original papers presenting tho Prooident?o memo mentioned only ASCII. and BCD as the codes which were considered; thuo? EBCDIC, the dominant and do facto standard, was not even consitlerod. I believe this ommisoion re?ected the dopth of the roaoarch into the conversion problema. Obviously. the Agency could go along v. ith the proposal for ?atandaros sake? but the working technicians would not take tho standard ooriouoly excoot through cacoosivo and unwiao coercion. .. Approved For ReleaSe 2002/01/29 CIA-RDP78-04723A000100150038-9 30 September 1968 MEMORANDUM :03: Chief. Informetion Processing - - SUBJECT Draft of NBS Instructions for Application of the Federal Standard ASCII Code n. clonal. Memo 13 Sep 68 b. NBS Memo 4 Sep 68 1. One of the problems that we see with the draft is the -. variety of interpretation that may be drawn from its language. depending upon. the reader?s role in the ADP world. From the standards enforcement viewpoint. there are references to ?Public Law 89-306. the ?President. and "1 July l969"--when ?all com- putters brought into the Federal Government inventory must have the capability to use ASCII. etc. "--which imply heavy handed enforcement. From the operations viewpoint. there is considerable language describing "evolutionary transition. and the authorization of departmonts and agencies to "waive: application of these guidelines" in certain situations. from which one might assume unpreasnred implementation. All of which leads one to wonder when and with what force application of these instructions is approP'l-iate. In addition to these differences in policy viewpoint. it has been quite evident from technical and management discussions generateti by the draft that ambiguom definition: are being drawn from name of the technical language in the paper. 2. In 1964. the 360 was offered with the internal options of EBCDIC or ASCII-3 (IBM's own transliteration of seven bit ASCII to eight bit bytes). So far as we can determine. no IBM 360's have been delivered which internally use ASCII-8. Because no standard eight bit code has been specified and approved by USASI (United States of America Standards Institute), EBCDIC has become the de ?facto industry standard for the internal logical Operation of computer systems. USASI has not npprovod a transliteration algorithm which would describe how seven bit byte ASCII appear: lagically within an eight hit'byta computer. Thin information in required for the development i '61" i - . a . EgApprovecl For Release 2002/01/29 CIA-RDP78-04723A000100150038-9 2 of software and programs which manipulate bit: rather than characters. With this in mind, i am. not sure what is meant in paragraph 7 which atatoa that all computers brought into the government "after 1 July 1968 must have the capability- to use ASCII. etc. Door this mean read in or write out only. or does it also imply a standard ASCII internal machine logic which can ho use-d for software development and the ?special application pragrams doacrihed above? If it mean: the latter-mud this interpretation can easily ho made-4hr impact will he never: on users and manufacturers. The numbar of manufacturers who will be prepared to deliver equipment which meet: this specifica- tion by 1 July 1969 in not clear now and this may force some relaxation of the proposed deadline. 3. The ASCII standard does not appear to present any serious problems in the interchange of information. Programs can be uucd to convert the input or output of a computer from EBCDIC to ASCII or ASCII to EBCDIG. In electrical transmission. .1 the computer controls for communication line: provides: automatic conversion of the eight bit bytoo in the computer to the acvcn hit bytes for transmission, and vice voraa. However. I believe that there are many subtle technical problems concerned with the data files. special programs. software and internal computer logic which will need to be solved before the Federal ADP standards program can reach the ultimate objective: of applying ASCII based code. media. and uqucnce Itandarda to internal filer. These problems will take a long time to resolve. Charles A. Briggs CHARLES A. BRIGGS Information Procter-ins Coordinator Salmon and Technology Distribution: Orig - Aden 2 - Registry - File Chrono STATINTL OSep68) i, 1 i . APPrWepl Ear.Release?2092191129 .- Approved For Release 2002/01/29 Next 1 Page(s) In Document Exempt Approved For Release 2002/01/29 CIA-RDP78-04723A000100150038-9