nls-restore
[Top] [All Lists]

Re: [nls-restore] Working out cloning logistics - part 2

To: nls-restore <nls-restore@chm.cim3.net>, nls-restore@chm.cim3.net
From: Philip Gust <gust@NouveauSystems.com>
Date: Sun, 20 Mar 2005 08:42:25 -0800
Message-id: <6.2.1.2.2.20050320084216.04108f58@mail.nouveausystems.com>
Rayleen,    (01)

Thanks for you retailed response to these questions. The information will 
be very useful not only for us during the preservation effort, but also as 
part of what we include in the curatorial layers to help people understand 
the system.    (02)

At 11:55 AM 3/19/2005, Raylene Pak wrote:
>4) What additional tasks are required once the archive made from this 
>backup is reloaded on a
>new system before we have a usable version of Augment running?
>
>My current thoughts are to leave the Augment/system configuration in such 
>a state that only the following will be required:
>.using emacs, modify a small number of TOPS20 files to incorporate the new 
>host's name, IP address and network connectivity information.
>.modify a file's name to include the new host's name.
>.run a script file to incorporate the new host's name into the Augment 
>executables.
>.use Augment to modify the journal configuration file to reflect hosting 
>journals on the new system.
>
>If the users want an operational SMTP deamon,  they will need to use emacs 
>to modify the host table to include the address of a host that is willing 
>to relay inter-host emails to their intended recipients.
>
>There may be one or two more steps, but this is all I have identified at 
>this time.
>
>I currently plan on setting up two Augment user account templates - one to 
>act as a programmer account and one to act as a "normal" user account 
>.  Creating new users is not difficult.  There will also be the operator 
>account with all the privileges it requires.  This account is not normally 
>used in conjunction with Augment; it is used to operate the emulated 
>DEC20.   Finally, there will be two "role" accounts that are used to 
>monitor the journal and journal search processes.
>There will be a generic ident system that all users will be members of and 
>a generic journal for users to experiment with.  I am assuming that Doug 
>will allow the Augment journal, purged of all private documents, to be 
>included in this release; however, this needs to be finalized with him.
>
>There are currently four Augment deamons configured to automatically start 
>up when the emulator is booted up.  They process inter- and intra-host 
>email and journal operations.  There are three other Augment deamons that 
>have been commented out of the start up run file; they deal with multiple 
>host journals and the suspense subsystem/application.
>
>7) Also on the 8th item, you mentioned at Doug's party that you rebuild 
>Augment relatively
>often.  It would be interesting to hear about what kinds of changes make 
>it necessary to rebuild
>Augment, what that process is like, whether it's incremental or requires 
>rebuilding the entire
>system, and how that impacts users of a "production" Augment system.
>
>Changes are made to Augment whenever Doug wants to modify or enhance 
>Augment.  Most recently, he wanted to add the capability to recognize URL 
>"links", "get" the text it specified and incorporate it into an Augment 
>file for display in the specified window.    Additionally, the 
>proliferation of spam has required frequent additions to the spam 
>filtering code in the Augment SMTP deamon.
>
>Augment source code is contained in Augment files.  At some point fairly 
>early in the NLS development process, Doug had the programmers using 
>Augment to develop Augment.
>
>Augment uses its own compilers and link/loader to create object and 
>executable files.  It has its own compiler-compiler to generate its 
>compilers.  It also has its own debugger.  That said, the source code does 
>contain small amounts of DEC10/20 assembly language and calls to the 
>TOPS20 operating system's interface procedures (jsys calls).
>
>The Augment Programs subsystem/application is used to compile all the 
>types of Augment source code/objects - e.g. subsytem backends and 
>grammars, sequence generators, straight L10 code, parse functions, 
>etc.  If a "program" is too large to reside in a single Augment file, a 
>TOPS20 script file is used in conjunction with the Augment link/loader 
>program (bldaug) to generate the appropriate type of executable from 
>multiple object files.
>
>Augment is comprised of two relatively large pieces of code; the frontend 
>(FE) and the backend (BE), a multitude of subsystems/applications and a 
>handful of deamons.
>
>The code for the FE and BE resides in multiple Augment files.  These files 
>are primarily located in the fesrc and besrc directories.  Most of the 
>code for these two modules is written in L10.  There is a small amount of 
>assembly language code and jsys calls included in these files.  When 
>changes are made to the code in one of these files, it is recompiled using 
>Programs subsystem.  Then, the appropriate link and load script in the 
>lib10 directory is used to generate a new version of the FE/BE.  The BE 
>was large enough to require an overlay executable.  Whenever the BE 
>executable is generated, its overlay must also be re-generated - and vice 
>versa.
>
>The Augment deamons also consist of multiple Augment files containing L10 
>code, assembly code and TOPS20 jsys calls.  Their source code files are 
>located in the utilsrc directory.  There are link and load scripts in the 
>lib10 directory used to generate the deamons' executables from the object 
>code.  Four of the Augment deamons must run in conjunction with BE executable.
>
>The Augment subsystems/applications consist of 2 - 4 seperate sections of 
>code - usually all located in one Augment file.  All Augment subsystems 
>have a grammar and backend code section.  The grammar is written in the 
>Command Meta Language (CML) and the backend code is written in L10, 
>assembly and TOPS20 jsys calls.  Some, but not all, subsystems also have 
>code sections for grammar parse functions and user profile features.  The 
>grammar parse functions are written in L10.  The user profile function is 
>written in its own language.  Each section of code is compiled 
>seperately.  If there is no change in the interface between the code 
>sections, it is not necessary to compile all of a subsystem's sections 
>when a change is made to just one section.  A couple of the subsystems' 
>backend code sections are too large to be stored in a single Augment 
>file.  In these instances that are build scripts to link the object files 
>to create a sin gle executable.
>
>Raylene
>  _________________________________________________________________
>Message Archives: http://chm.cim3.net/forum/nls-restore/
>Shared Files: http://chm.cim3.net/file/work/project/nls-restore/
>Community Portal: http://www.computerhistory.org/
>To Post: mailto:nls-restore@chm.cim3.net
>Community Wiki: http://chm.cim3.net/wiki/    (03)


Philip Gust
Nouveau Systems, Inc.    (04)

phone: +1 650 961-7992
fax:   +1 520 843-7217    (05)


mailto: gust@NouveauSystems.com     (06)


 _________________________________________________________________
Message Archives: http://chm.cim3.net/forum/nls-restore/
Shared Files: http://chm.cim3.net/file/work/project/nls-restore/
Community Portal: http://www.computerhistory.org/  
To Post: mailto:nls-restore@chm.cim3.net
Community Wiki: http://chm.cim3.net/wiki/    (07)
<Prev in Thread] Current Thread [Next in Thread>