Unix System Laboratories: Difference between revisions

Revision as of 01:31, 3 August 2020

Revision as of 12:11, 20 November 2020

Line 44:

== Origins as subsidiary of AT&T ==

{{see also|History of Unix|Unix Wars}}

AT&T announced the creation of the [[UNIX Software Operation]] (USO) – a separate and distinct AT&T business unit responsible for the development, marketing, and licensing of UNIX System V software – in January 1989.<ref name="signals-91">{{cite news | url=https://books.google.com/books?id=UY0-AQAAIAAJ | title=... | magazine=Signals | date=1991 | pages=61–66}}</ref> This was done, as a subsequent press release stated, "in order to separate AT&T's UNIX System source code business from its computer systems business,"<ref name="pr-rosetta"/> the latter a reference to [[AT&T Computer Systems]]. USO included the AT&T Data Systems Group organizations responsible for UNIX product planning and management, licensing, and marketing.<ref name="pr-uso"/> [[Peter J. Weinberger]] was named chief scientist of USO while also retaining his job in the computing science research center at [[Bell Labs]]; no other Bell Labs assets were transferred to USO.<ref name="pr-uso">{{cite press release | url=http://tech-insider.org/unix/research/1989/0104.html | title=AT&T Names President Of Unix Software Operation | publisher=PR Newswire | date=January 4, 1989}}</ref> The head of USO was Larry Dooling, who had been a vice-president in sales and marketing in the AT&T Data Systems Group.<ref name="pr-uso"/>

Unlike the original Unix work, which had been done in the Bell Labs facility in [[Murray Hill, New Jersey|Murray Hill]], USO and the commercialization work was done a few miles away in [[Summit, New Jersey]],<ref name="negus"/><ref name="lat-novell"/>

Line 53:

UNIX System Laboratories, Inc., came into being as a separate subsidiary of AT&T in November 1989 and was assigned all U.S.-based AT&T Unix and USO assets.<ref name="pr-rosetta"/>

However USO continued to operate as USO until June 1990, when the reincorporation of AT&T's European and Asian Unix business operations as wholly owned subsidiaries of USL was completed.<ref name="signals-91"/> At that point the UNIX Software Operation was publicly rebranded as UNIX System Laboratories.<ref name="pr-rosetta">{{cite press release | url=http://tech-insider.org/unix/research/1990/0625.html | title=AT&T Renames Unix Software Unix System Laboratories | publisher=PR Newsire | date=June 25, 1990}}</ref>

Again, a point of emphasis was to separate the Unix-based business from AT&T's hardware-based business.<ref>{{cite news | url=https://books.google.com/books?id=0bqdMvDMv74C&pg=PA10&lpg=PA10 | title= AT&T's Unix unit spun off | first=Johanna | last=Ambrosio | newspaper=Computerworld | date=July 23, 1990 | page=10}}</ref> The subsidiaries were known as UNIX System Laboratories Europe, Ltd., sited near [[Ealing Broadway]] in [[London]], and UNIX System Laboratories Pacific, Ltd., located in [[Shiba, Tokyo]].<ref>{{cite book | url=https://books.google.com/books?id=xe-8gZ6FoIMC&pg=PA39 | title=AUUG Conference Proceedings 1992 | publisher=Australian Open Systems Users Group | date=Summer 1992 | page=39}}</ref>

Dooling was named the initial president of USL, continuing from his position at USO.<ref name="pr-rosetta"/>

Line 67:

Next USL engaged in an especially arduous effort into trying to satisfy the requirements of the [[National Computer Security Center]]'s [[Trusted Computer System Evaluation Criteria]] ("Orange Book") to the B2 level.<ref name="cw-sec">{{cite news | url=https://books.google.com/books?id=a6P8XXI-B8cC&pg=PA32 | title=AT&T Unix to gain enhanced security | first=Johanna | last=Ambrosio | newspaper=Computerworld | date=October 1, 1990 | page=32}}</ref> This manifested itself in System V Release 4.1 ES (Enhanced Security), which also included generally useful features such as support for dynamic loading of kernel modules.<ref>{{cite news |title=Unix International reviews the Unix System V.4 story so far |first=William |last=Fellows |date= August 13, 1992 |work=Computergram International |publisher=Computer Business Review |url=<!--blacklisted http://www.cbronline.com/news/unix_international_reviews_the_unix_system_v4_story_so_far --> }}</ref>

Following that, USL worked on [[UNIX System V#SVR4.2 / UnixWare|System V Release 4.2]], which was released in June 1992.<ref name="iw-42">{{cite news | url=https://books.google.com/books?id=7D0EAAAAMBAJ&pg=PA1 | title=Univel targets UnixWare release for early fall | firstfirst1=Vance | lastlast1=McCarthy | first2=Cate | last2=Corcoran | newspaper=InfoWorld | date=May 11, 1992 | pages=1, 103}}</ref> ''[[InfoWorld]]'' characterized this effort as "at the core of an assault on the enterprise networking market," with a modular architecture that stressed improved support for enterprise- and network-level administration, drivers for both [[Token Ring]] and [[Ethernet]], and a greater ability to run on low-end machine configurations.<ref name="iw-42"/>

Multiprocessing became the focus of the final USL-based OEM release of System V, which was Release 4.2MP, released in December 1993.<ref name="timeline"/>

USL continued the publication of an early Unix standard, the [[System V Interface Definition]] (SVID).<ref name="advanced"/> Moreover, the SVID became one of the bases for the more important, vendor-independent [[POSIX]] standard for Unix, which System V Release 4 releases also conformed to, as well as the later [[Single UNIX Specification]].<ref name="advanced">{{cite book | <!-- https://books.google.com/books?id=_kpsAQAAQBAJ&pg=PA33 --> | title=Advanced Programming in the UNIX Environment | firstfirst1=W. Richard | lastlast1=Stevens | first2=Stephen A. | last2=Rago | publisher=Addison-Wesley | edition=Third | ___location=Upper Saddle River, New Jersey | date=2013 | pages=32–33}}</ref>

USL produced many books documenting various aspects of Unix System V.<ref>{{cite web | url=http://worldcat.org/identities/lccn-n91048302/ | title=Identities: UNIX System Laboratories | publisher=WorldCat | accessdate=May 6, 2018}} WorldCat lists "122 works in 297 publications in 1 language and 1,849 library holdings" from USL.</ref>

Line 88:

Another unit within USL, called the Open Solutions Software business unit and headed by Joel A. Appelbaum, was responsible for other system software that in some way worked in conjunction with Unix.<ref name="pr-rosetta"/>

The [[Tuxedo (software)|Tuxedo transaction processing middleware]] had also been transferred from elsewhere in AT&T to USL. It had originated as the [[Loop maintenance operations system|Loop Maintenance Operations System]] (LMOS) followed by the Unix Transaction System (UNITS) and was used for projects internally within AT&T. It was then renamed by USL and, as Release 4.0 of what was now called Tuxedo, in 1989 was offered for the first time as a commercial product.<ref>{{cite book | url=https://archive.org/details/zdanh_test_031_tuxedosystemsoft00juan| url-access=registration<!-- https://books.google.com/books?id=Of6BdiZu9lUC&pg=PT27--> | firstfirst1=Juan M. | lastlast1=Andrade | first2=Mark T. | last2=Carges | first3=Terence J. | last3=Dwyer | first4= Stephen D. | last4=Felts | title=The TUXEDO System: Software for Constructing and Managing Distributed Business Applications | publisher=Addison-Wesley | ___location=Reading, Massachusetts | year=1996 | pages=xxxiii–xxxvi}}</ref>

USL also developed and marketed the OSI Communications Platform, which was an implementation of the [[OSI model|OSI protocols]] for Unix-based networking.<ref name="nw-knowhow">{{cite news | url=https://books.google.com/books?id=cBAEAAAAMBAJ&pg=PA9 | title=Novell buys into AT&T unit to gain Unix net know-how | first=Ellen | last=Messmer | newspaper=Network World | date=April 15, 1991 | page=9}}</ref>

Line 95:

{{see also|C++#History}}

There was also a languages department at Unix System Laboratories,<!-- https://books.google.com/books?id=9yMQAQAAMAAJ Dr. Dobb's Journal of Software Tools ... is member of the languages department at Unix System Laboratories and is the editor of the Standard for the C++ Standardization committee. --> which was responsible for the [[C language]] compiler and development tools used to build Unix.<ref name="pr-uso"/> Moreover, it was responsible for commercial sales related to the C++ language, including development tools such as the [[Cfront]] compiler that had come from AT&T.<ref name="d-and-e"/>

Indeed, the paper describing one of the first implementations of automatic instantiation of [[Template (C++)|C++ templates]] in a C++ compiler had as lead author an engineer associated with Unix System Laboratories.<ref>{{cite journal | firstfirst1=Glen | lastlast1=McCluskey | first2=Robert B. | last2=Murray | title=Template Instantiation For C++ | journal=SIGPLAN Notices | volume=27 | issue=12 | date=December 1992 | pages=47–56| doi=10.1145/142181.142195 | s2cid=27330199 }}</ref><ref name="m-and-c"/> And [[Margaret A. Ellis]], co-author with C++ creator [[Bjarne Stroustrup]] of ''The Annotated C++ Reference Manual'', an important publication in the history of the language, was a USL software engineer.<ref name="d-and-e"/>

[[Image:Office within the Unix System Laboratories building in Summit, New Jersey--March 1994.jpg|thumb|left|A software developer working in the Summit building]]

USL also continued the development of, and attempted to market, C++ Standard Components,

an early instance of a C++ software foundation library that supported [[Container (abstract data type)|container classes]] and various other [[computer science]]-based functionality such as [[finite-state machine]]s, [[Graph (abstract data type)|graph]]s, and [[regular expression]]s.<ref name="udk-sc"/><ref name="report"/> The Standard Components originated in conjunction with early developments in the C++ language in Bell Labs and became widely used internally within AT&T,<ref name="d-and-e"/> by one estimate being used in hundreds of projects.<ref>{{cite book | <!-- url=https://books.google.com/books?id=AhI7BAAAQBAJ&pg=PA146 --> | firstfirst1=M. Afshar | lastlast1=Alam | first2=Tendai | last2=Padenga | title=Application Software Reengineering | publisher=Pearson | ___location=Delhi | date=2010 | page=146}}</ref>

They represented an effort among early library writers there to design reusable code using C++ idioms.<ref name="m-and-c">{{cite book | firstfirst1=Martin D. | lastlast1=Carroll | first2=Margaret A. | last2=Ellis | title=Designing and Coding Reusable C++ | publisher=Addison-Wesley | ___location=Reading, Massachusetts | date=1995 | pages=x, 231}}</ref> Unlike its other offerings, which were sold to OEM vendors and resellers, here USL sold to end users.<ref name="report">{{cite news | title=Product Review: The USL C++ Standard Components Release 2 (end user package) | first=Bill | last=Leggett | magazine=C++ Report | date=June 1992 | pages=69–73}}</ref> The initial release of USL C++ Standard Components to the general computing industry was labelled as Release 2.0 and occurred in 1991;<ref name="udk-hist"/> it suffered from an awkward mechanism to get around the lack of templates in the container classes.<ref name="report"/> That was followed by Release 3.0, which added support for templates, in 1992.<ref name="udk-hist">{{cite web | url=http://osr600doc.sco.com/en/SDK_c++/_Product_History.html | title=Product History | publisher=The SCO Group | date=June 2, 2005 | accessdate=April 28, 2018}}</ref> <!-- USL C++ Standard Components Release 3.0 Programmer's Reference - could include cover of that C++ SC manual as fair use if I still have it somewhere -->

Some within USL believed that C++ Standard Components could become a language standard as well as a significant source of revenue,<ref name="d-and-e"/> but it had trouble gaining traction outside of AT&T. Stroustrup would later describe these goals as "a misguided belief".<ref name="d-and-e">{{cite book | first=Bjarne | last=Stroustrup | title=The Design and Evolution of C++ | publisher=Addison-Wesley | ___location=Reading, Massachusetts | date=1994 | pages=124–125, 126–127, 184 | bibcode=1994dec..book.....S }}</ref><!-- see the HOPL II/III? paper if I have it ... looks like nothing in either --> In any case, all such libraries were soon eclipsed by the radically different [[Standard Template Library]] (STL), which became the standardized foundation library for the C++ language.<ref>{{cite conference | contribution=Evolving a language in and for the real world: C++ 1991-2006 | first=Bjarne | last=Stroustrup | title=Proceedings of the third ACM SIGPLAN conference on history of programming languages | date=June 2007 | publisher=ACM | pages=4-8–4-11 }}</ref> As it happens, one of the Standard Components, array_alg, was designed by the creator of STL, [[Alex Stepanov]], and can be considered an early predecessor of STL.<ref>{{cite web | url=http://stepanovpapers.com/ | title=Alexander A. Stepanov | publisher=stepanovpapers.com | date=October 6, 2016 | accessdate=December 15, 2017}} and {{cite web | url=http://osr600doc.sco.com/en/SDK_c++/_No_More_Array_Errors_Part_II_-_.html | title=No More Array Errors (Part II) - Array_alg(C++) | publisher=The SCO Group | date=June 2, 2005 | accessdate=April 28, 2018}}</ref>