soft-eng@MITRE.ARPA (Alok Nigam) (09/09/88)
Soft-Eng Digest Wed, 7 Sep 88 V: Issue 29
Today's Topics:
call for papers
looking for unix based CASE tools (2 msgs)
Marketing software (2 msgs)
rapid prototyping
Re: Shh, I'm working
Reverse engineering, design assessment support
software metrics
Version control across machines (2 msgs)
WANTED: ideas, methods and tools for automated tests of C,.. pgms
----------------------------------------------------------------------
Date: 5 Sep 88 23:13:59 GMT
From: oliveb!intelca!mipos3!omepd!psu-cs!warren@ames.arpa (Warren Harrison)
Subject: call for papers
Call for Papers
for a Special Issue of
The Journal of Systems and Software
on
"Using Software Metrics"
Software metrics have been a popular topic among both researchers
and practitioners. However, little work has been published
describing how metrics can be applied in the development, testing
and maintenance of computer software. This special issue will
focus on the topic of using software metrics as opposed to
simply studying them.
Individuals from both academia and industry who are currently
using software metrics in the classroom or in the field are
encouraged to submit a paper describing their experiences. Papers
should address three points:
(1) Which metrics are being used?
(2) How are they being applied?
(3) How successful have they been?
All papers will be refereed by a panel of researchers and
practitioners. Deadline for submissions is January 1, 1989. For
more information, or to submit a paper, contact:
Warren Harrison
Department of Computer Science
Portland State University
PO Box 751
Portland, OR 97207
------------------------------
Date: 2 Sep 88 21:37:49 GMT
From: ihnp4!arrow!dcb@bloom-beacon.mit.edu (DC Becker)
Subject: looking for unix based CASE tools
Am looking for any pointers to products now or soon to be available
in the way of CASE tools, running under unix. I know HP has
the TEAMWORK package that runs on their HP-UX. They get real
picky about running it on their boxes with their OS though.
There is also a pile of products out there that run on PCs and
workstations of various flavors. Has anyone successfully integrated a
development environment with PCs and a large unix host environment?
------------------------------
Date: 4 Sep 88 04:28:14 GMT
From: hpl-opus!hpccc!hp-sde!bd@hplabs.hp.com (Bob Desinger)
Subject: looking for unix based CASE tools
> Am looking for any pointers to products now or soon to be available
> in the way of CASE tools, running under unix.
Ask your librarian to do a literature search for you. If you ask for
anything with the keyword "CASE" or the phrase "Computer Aided
Software Engineering" in it, and limit the search to 1988 or 1987,
you will come up with literally hundreds of references.
To narrow the search, limit it down to 1988, or to articles mentioning
only specific companies (such as HP or Hewlett-Packard, to use your
example), or to specific products (such as Teamwork). Or you can
limit the type of source material to product announcements if you want
the broad overview, or technical papers if you want the philosophy
behind the products. You may want to request just the titles of the
articles, then ask your librarian to pull the full text on only the
ones that look interesting.
------------------------------
Date: 26 Aug 88 05:49:42 GMT
From: tektronix!reed!psu-cs!warren@bloom-beacon.mit.edu (Warren Harrison)
Subject: Marketing software
It all depends what you want out of the deal. There are lots of ways to get
free or low cost publicity for your product. I'll mention them from most
effective to least effective - I've done all these for my software, so I
can only relate my experiences, but I think they're representative.
(1) Product Reviews in a magazine. This is by far the most effective way
to get your product known (if it is a good review). The approach I
have found the most useful (if your product is unique, which I think
yours is) is to send a press release to the editors of every magazine
you can think of (see #2). If the product looks interesting, they'll
call you about a review copy (sometimes). You're safer of course
if you just send out free copies to magazine editors and/or reviewers.
Probably 1% of the recepients will even look at your software, but the
last time PC-Week reviewed one of my programs, we probably got 100 or
so orders out of it (consider a normal month for us is 30-50 orders).
If you get your documentation in lots of 500 at a time (the most
cost effective), it probably costs less than $5-$10 to send out a
review copy. What's even better is that after a magazine has reviewed
your product you can send out copies of the review with your literature
when someone calls for info. This adds credibility and helps make up
for the crummy brochure you're bound to come up with during the first
year of business (see #3).
(2) Press releases in magazines. Write up a press release and send a
copy (or three) to every name you can find inside the cover of
every computer magazine you can think of. These folks get hundreds
(really) of press releases every month, so don't expect everyone
to run your product announcement, but if one or two do, it could
result in some sales. More importantly, it continues to add to the
credibility of your product. The thing most buyers are afraid of is
getting ripped off. This helps show that your product really exists
and they're not sending their money down a black hole. Go to a
library at a nearby college and check out their journalism texts.
They can help you write a press release.
(3) Low cost ads in magazines. Computer Language and Dr. Dobbs both have
"Programmers Marketplace" sections as do Programmer's Journal,
Micro Cornucopia, etc. Rates can run from $100 to $400 a month for
a 1/9 page ad. They're not as effective as a full page color as
on the cover, but do you really want to spend $5000 for a one shot ad?
As a bonus, most of these places will typeset your ad for you ...
doing this for the full page ads can cost almost as much as the ad
itself. We commonly get about 40-50 reader service inquiries a month
off each of our ads. Remember, you still have to sell these people
once they make an inquiry, so you'll have to have some brochures or
other literature on hand to send to people when they request more
information. Again, if you have a copy of a review from a known
magazine to send with the literature, it really helps.
As to start up cost you can figure documentation and binders for the
documentation. Brochures, and money for some low cost ads, as well as
actually setting up the brochurs, documentation and ads (if the mag won't
typeset them for you). The costs can vary quite a bit, depending on how
much documentation, etc. you need. You will also need an actual business
phone line (your residence line won't hack it). When someone sees a
product announcement or review that lists your company's name but not its
phone number, you want a potential customer to be able to find it by calling
information. Also, you want someone to answer the phone during the day.
You can use an answering machine once in a while, but if a person gets a
machine day after day, every time they call, they can't help but suspect
your technical support will be a bit weak.
Feel free to send me mail if you have specific questions. I know you can
do it, because I have. I'll never get rich off it, but it does tend to
pay for my toys (like my Porsche and my 386). Just make sure you have
another job to live on.
By the way, I tried it via the shareware route first and have concluded that
unless you have a product that everyone that owns a PC would use, you're
wasting your time. Less than 5% of the users of shareware actually pay
the full registration price, but many of the overhead costs are still
there, like legal fees, banking charges, postage, etc.
------------------------------
Date: Fri, 2 Sep 88 15:08:43 BST
From: Gordon Howell <mcvax!hci.hw.ac.uk!gordon@uunet.UU.NET>
Subject: Marketing software
> Now my problem. Everyone tells me that to market this software will
>require something like $500,000 up front for publicity. I don't have this
>kind of money and don't have a line of venture capitalists beating a
>path to my door. Isn't it possible to bootstrap myself. By this I mean,
...
This should be a question of general interest to us all. How many
software engineers out there *haven't* either written some software or
thought about writing software for fun and profit? How many of you
are also shark-class businessmen?
I would like to see discussion on this subject, but I am not sure this
is the appropriate forum. Is there any group dedicated to the
business of software (as opposed to "business software"!)? If not,
are there people interested in starting such a discussion group?
For example, what you may yearn for is a consortium similar to an
insurance group among cooperating software engineers. Not a lawyer, I
wouldn't have a clue how to go about such a scheme (or other business
arrangements)...
------------------------------
Date: Tue, 30 Aug 88 14:42:11 PDT
From: PAAAAAR%CALSTATE.BITNET@cunyvm.cuny.edu
Subject: rapid prototyping
>I use Smalltalk/V from Digitalk for rapid prototyping and systems
>simulation. But its not what I would call rapid prototyping in the
>traditional sense
It helps to have a name for different kinds of prototypes. There was some
correspondence in SIGSOFT "Software Engineering News" a while back that
deefined a few useful differences - breadboard, mock-up, scale model...
>traditional sense ( quickly building a system which has the functionality
>of the target system but not (perhaps) meeting the performance or
>limits criteria ).
The above is a "mock-up" - it looks like the final version
but you can't "really" drive it anywhere. Specifically designed for
a show and tell session with users/clients/vips. A Mockup looks
good but can't do the whole task.
>So, I simulate the network and simulate the devices.
>Then I simulate the "relevant" software around and inbetween.
[...]
>this type of prototyping is extremely
>valuable to me: Message protocols are defined very quickly (Smalltalk
>excells at message passing) and error detection, reporting, and correction
>are quickly developed.
>
>I quess I really use Smalltalk/V as a system design tool. Still...
>I can show my boss... "this is how its going to work" ... But it's
>not something I'd show a customer. (Why, Ed? Hum, I'm not sure.)
This is a "Bread-board" prototype.
A breadboard was defined to be a
messy, homemade, hacked-up collection of objects that communicate
by easily modified channels. Specifically for the engineer and perhaps
his/her boss. *Not* for the user/client/vip to see.
A Scale-model is prototype version that has the full functionallity but is
small scale - this means it does not have all the complete data base
available, or perhaps, not the complete RAM thats need, or the full
knowledge base. It's useful for engineers and clients/user/vips and bosses.
You can also develop - "executable blueprints" and "Performance models" that
make it more likely that you don't get egg on your terminal I guess?
Are there any other varieties of prototype out there?
------------------------------
Date: 5 Sep 88 02:51:12 GMT
From: att!chinet!mcdchg!clyde!watmath!watcgl!electrohome@bloom-beacon.mit.edu (Joe Buccino)
Subject: Re: Shh, I'm working
At my previous job, people had shared cubicles, at which all but "official"
business was performed (often crowded in by stacks of paper or unwanted
equipment). The great debate over the past 6 months was over how to
distribute the few window cubicles (about 18 of 40) amongst the different
teams and personnel. Privacy was non-existent. Background noise was
beyond what I would consider comfortable in most cases; I had a pair of
earplugs that were essential to being able to think creatively. And, to
address the initial issue, there was no room or area where you could go to
get away from all of the commotion; instead, this company that prided itself
on being a "billion-dollar-a-year" company took all of their remaining space
and built more offices for more managers while cramming the plebes closer and
closer together. That's one of the many reasons why I was able to
address this as "my previous job".
------------------------------
Date: Fri, 26 Aug 88 12:23:21 -0400
From: howell%community-chest.mitre.org@gateway.mitre.org
Subject: Reverse engineering, design assessment support
Several CASE vendors now advertise support for "reverse engineering" of
source code to some design representation; I'm interested in any
experiences with any of these capabilities. I'd appreciate any
information about vendors that are supporting reverse engineering and
any "consumer reports" on them.
I've also seen a description of a tool called INSPECTOR, from a company
named Language Technology. It is a "quality assurance tool based
on scientific measurement of COBOL quality and maintainability" (from a
CASE seminar announcement). Any information on this tool and others
like it (particularly for Ada) would be appreciated.
------------------------------
Date: 3 Sep 88 22:07:26 GMT
From: littlei!omepd!psu-cs!warren@uunet.uu.net (Warren Harrison)
Subject: software metrics
I will be giving a workshop on software metrics in the near future, and
would like to learn of any commercial software metrics packages that are
available so I can describe them in my lecture. I'd appreciate the name
and address of the company that sells them as well as your experiences
if you have had a chance to use one.
------------------------------
Date: 1 Sep 88 16:07:51 GMT
From: vsi!sullivan@uunet.uu.net (Michael T Sullivan)
Subject: Version control across machines
Now that we're _finally_ getting into SCCS here, things are getting more
organized as far as version control. However, when we port something
to another machine, what do you do? What I'm asking is, are there any
references on how to keep version control between machines manageable?
If there aren't, how do you people out there do it? With all these Unix
boxes around and plenty of people doing this sort of thing, I'm hoping
somebody has made it less of a nightmare than it has been for us in the past.
------------------------------
Date: 3 Sep 88 05:28:18 GMT
From: bsu-cs!dhesi@iuvax.cs.indiana.edu (Rahul Dhesi)
Subject: Version control across machines
>What I'm asking is, are there any
>references on how to keep version control between machines manageable?
I don't know how they do it, but here's one work-around.
Keep your master versions on a single machine that we will call the
"home" machine. Transfer them to any other machine, but once a day (or
once a week, etc.) move everything back to the home machine and check
it in as a new version in an appropriate branch. Software development
continues on the other machines and on those machines you could
continue to do version control locally. Several quick revisions on a
remote machine might correspond to only one revision on the home
machine.
------------------------------
Date: 25 Aug 88 11:51:43 GMT
From: mcvax!enea!log-hb!janne@uunet.uu.net (Jan Skarvall)
Subject: WANTED: ideas, methods and tools for automated tests of C,.. pgms
I need a way to make testing of software an easy task, so easy that I'll do
it every time before delivery of a software package and even after each
major change in that software. The software package can be written in C or
Ada, but that should not matter very much.
This, I am sure, has been done before by making use of Unix-make and possibly
also with other tools like SPMS (which uses make).
WANTED: 1) I would like to know if someone has put a little more thought in
how to use make for this purpose.
2) I also would like to know if there are more sophisticated tools
existing or on the way.
3) References to books, articles etc. on this topic.
4) SPMS for Sun.
------------------------------
End of Soft-Eng Digest
******************************