SD West '95? Opinions?
ro...@crl.com
last year? It seemed that everyone was hyped at the '94 show about Visual C++
coming out as a standardized, cross-platform software development tool. But,
with the subsequent reality that set in, and the realization that it would be
quite a bit of time (to whit, a couple of months ago) before VC++ 2.0 actually
hit the streets on more than one platform, it seems that people are showing a
bit more jaded behavior to MS product realeases. I didn't see droves of
individuals waiting in line to see MacWarp (aka Win96). I saw more people
sitting in on the OpenDoc demos, to be truthful. It might just be the last day
blues, but it really didn't seem like people are as hyped about Win95 as they
were about VC.
I also caught a lot of grumbling at the advanced VC lecture concerning small-
scale improvements to the VC IDE (things like rewind, break on change of value,
ability to edit accelerators, macro recording). This all sounds like 'cute
stuff' as far as the editor & debuggers are concerned, but doesn't really
address compiler performance, or IDE flexibility. As an example (please hold
your flame, VC'ers) from WorkFrame/2 2.1:
Projects are desktop objects. They are subclassed from directory folders,
and are represented as standalone or composite projects, taking the
appearance of directories containing source files. Composite projects can
contain sub-projects within them, thus giving the application developer the
ability to subdivide the project into manageable chunks (a UI sub-project,
a database engine sub-project, etc. On the composite project level, you
specify the standard methods and operators for given file types (for code
files, you select the particular compiler you need, for object files, the
appropriate linker, etc). You can also create methods to pre-compile files
of a particular type (i.e., run SQLPREP on embedded SQL .CSC files, SOMC on
SOM .IDL files, etc). You also select what editor to use when
double-clicking on a code file.
There's a couple of things to note here:
1. Nobody says what kind of editor you have to use.
2. Nobody says what kind of compiler you have to use, either in the way of
manufacturer *or* language.
3. There's no limit to the nesting available in your project (aside from
disk space, of course).
4. If you have code files requiring special action (.ASM, .CSC, .IDL,
whatever), you can run an appropriate pre-compiler on them, also
dictated only by your preference.
What we were able to see of the incremental build of the VC IDE (the one that
will be released to subscription members), none of these abilities are present.
Everything is still geared to use MS's compiler. The only way to run an external
pre-compiler on special-case files is to create a 'make' batch file and run the
batch file as the build command in an external make file (thereby losing all of
the appreciable flexibility of VC-native make file project management). You can
sub-divide your project into different areas, but the nesting level is only one
deep, and project management is much more clumsy (cascading notebooks of
accumulating/overriden option settings that you really had to dig to find). I
think I would be much more reassured in the approach to software development
that stresses parts development vs application development, the way that M$ has
said the future of software development will run (and IBM has taken to heart in
their development tools), if M$ would take the first step: namely, in *their*
development tools. I don't see that happening, though, and that leaves me
questioning M$'s posture as a 'do as I say, not as I do' affectation, and one
that only invites distrust and suspicion.
On other fronts:
Internationalization and localization of software applications.
It seems that a good number of companies have come a long way in taking an
overall approach to distributing applications to foreign markets. They stressed
first assessing the ability to generalize language-specific portions of your
applications before they would even consider localizing the application to a
given country. This is a big step from a couple of years back, where the most
you could hope for was a straight, across-the-board translation of your message
strings...
Hot product of this show: Delphi. Looks like Borland's betting the bank on this
one. Personally, I hope it does well. It seems like a really flexible RAD tool.
Unfortunately, I don't have much confidence in Borland weathering much more
abuse in the marketplace. Also, all of my projects are in C++ and need to be
updated *yesterday*. I don't have time to port to pseudo-Pascal, at least not
right now.
Database products: The consensus is in. ODBC is a neat tool for getting DB info
from one format to another, but its response time compared to close-coupled
database engines just isn't there. Neat toy, if you can afford it: the
S-Designor database modeling tool. Invaluable for setting up generalized,
scripted database structures that can subsequently be used to create working
databases in a wide variety of database sources using, yes, ODBC. The only catch
is, same as ODBC, you need the driver to create/manipulate the database.
Product warning: Version control systems, specifically Versions 1.1 from
StarBase. When choosing a version control system, look for one that's capable of
doing branches and merges. There were several ported from Unix->Windows that
did. Versions 1.1 is a cost-effective solution for Windows users requiring
revision tracking software, but does not have branching/merging capabilities.
The drawback to this is that you can't have more than one person working on a
given source file at any given time...you have no way of integrating independent
changes into a final composite file. This feature has saved my co-workers and I
from several catastrophes involving code overlap. Unfortunately, this was in MS
Delta, and that's being phased out in favor of SourceSafe. The jury's still out
as far as the decision to change over is concerned. If any of you have
glowing/horror stories about revision control software, I would greatly love to
hear them.
Cross-platform tools: OpenDoc. In Alpha for AIX, Mac, OS/2 and Windows, but
demonstrable on all platforms in a trade show environment nonetheless. Design an
editor and viewer in parallel (the viewer is just the editor without change and
save methods) for your given document type. Design them as 'part handlers': only
enough code to handle displaying, printing, and editing for that given format,
and you have instant portability to all the mentioned platforms. Includes OLE
compatibility, i.e., you can embed an OLE object in an OpenDoc file, and
vice-versa. The advantage is that the OpenDoc parts handlers have everything
they need to act on their part without having to call up an associated
application to expose the edit methods for that part. Also, the part handlers
are re-entrant, so only one is present in memory for a given part type,
regardless of how many objects of that part type are present in a document.
Anybody else out there get the chance to take in the show? What did you think?
Roby Gamboa
sder...@sensemedia.net
>Is it just my imagination, or did SD West seem quite a bit more subdued than
>last year?
MS's booths were very quiet. Deathly quiet was what I heard
as I walked by.
IBM's booths seemed equally quiet to me though.
>
>Hot product of this show: Delphi. Looks like Borland's betting the bank on this
>one. Personally, I hope it does well. It seems like a really flexible RAD tool.
>Unfortunately, I don't have much confidence in Borland weathering much more
>abuse in the marketplace. Also, all of my projects are in C++ and need to be
>updated *yesterday*. I don't have time to port to pseudo-Pascal, at least not
>right now.
I went Wednesday and Borland was really hopping for the 4
hrs I was on the floor. Delphi looks very cool. I'm going
to have to give it a try.
I don't think Borland is betting the bank on it though.
Rumors of whats going to be inside BC5.0 for Windoz/NT/DOS32
sound very very nice.....
>as far as the decision to change over is concerned. If any
of you have
>glowing/horror stories about revision control software, I would greatly love to
>hear them.
I had to use SourceSafe on my last contract and hated it. I
like to use commandline vcs tools and SS left about %90 of
the api out of their commandline tools. If your going to do
anything except checkout or checkin a few files you must run
the windows hosted app. And its a DOG and flacky...
RCS is still king in my book, on all platforms. I got a demo
of MKS's won/dos latest and greatest. When I try it out
I'll post my very subjective opinion....
BTW MKS's RCS is completely compatible with the shareware
OS2 and unix flavors....
>Anybody else out there get the chance to take in the show? What did you think?
I went, I saw, I left... I'm not sure it was worth the one
day of lost hours though..
--------------------------------------------------------
Scott Derrick | Yahoo Engineering
sder...@sensemedia.net |
sc...@cats.ucsc.edu |
(408) 335-7373 | "Make it so!"
--------------------------------------------------------
bill...@taligent.com
>last year?
I thought so too. Seemed like the same old "Visual Basic/C++ and Clones World"
show that it was last year, but now people are tiring of that.
>Cross-platform tools: OpenDoc. In Alpha for AIX, Mac, OS/2 and Windows, but
>demonstrable on all platforms in a trade show environment nonetheless. Design an
>editor and viewer in parallel (the viewer is just the editor without change and
>save methods) for your given document type. Design them as 'part handlers': only
>enough code to handle displaying, printing, and editing for that given format,
>and you have instant portability to all the mentioned platforms.
Instant? I suppose you first have to somehow get the underlying APIs you're
using to do this displaying, printing, and editing ported to all these platforms.
It seems that you can maybe recompile the portion of your code that uses the
OpenDoc APIs, but this is a small percentage. Or is there more to it?
Bill Law
Marvin Lichtenthal
it was down in Silicon Valley. I only went for Exhibits and some
of the no-fee sessions.
At times the exhibit hall looked very un crowded.
Part of the problem I think is that the Windows 95.11 fiasco has
hurt the industry real bad. I think a lot of companies are hurting
since they have been doublecrossed by MS and have no product to
sell until 1996. Delphi seemed to be the highlight product, and
Symantic new compiler sounded a little promising.
The Symantic party had the best food. Watcom had no announcement
session, their party was ran out of food rather quickly.
However they more than made up for it with the great free
software. Everyone got a special
SD95 SQL for WIndows 4.0 package. I also won a copy of VXREXX 2.1 and
Watcom C/C++ 10.0. Seem to be giving a lot of prices.
Forgot to mention Phillip gave all attendees a copy of SideKick 2.0.
Borland gave all a CDROM version of Delphi.
Intel had some nice exhibits and the Apple booth was showing off
PowerMac versus Intel benchmarks.
There were fewer drawings for prices than in prior years. This is
where you put your card in a jar and they ...
Looking forward to hear what others thought.
--
Marvin Lichtenthal
mar...@netcom.com
Richard Smith
Here is my take on SD '95:
Here is what happened out at SD '95. At this year's
I was disappointed that not alot of new products were
introduced at the show. Because Windows
95 isn't shipping, it seems that many development tools
companies are currently in a holding pattern.
The one new product that was introduced at the show was
Borland's Delphi. Delphi is basically a Visual Pascal
compiler with a built-in database. It comes in two
flavors, a desktop version and a client/server version.
The desktop version competes with Visual Basic and the
client server version competes with PowerBuilder. The
product demos very well and Borland was able to keep
large crowds around their booth for all 3 days of the
tradeshow.
Delphi addresses a number of problems that people have
been complaining about in Visual Basic over the years.
First of all, Delphi includes a true compiler unlike
Visual Basic which has a Pcode interpreter. That means
that Delphi applications on average run much
faster than Visual Basic applications. Secondly, a
Delphi application doesn't need a bunch on extra files
surrounding the application because the run-time
system is linked with the applications. This feature
makes it easier and cleaner to redistribute applications.
Finally, Delphi allows a programmer to write controls
in the same language as applications. In VB, programmers
have to switch to C or C++ to write controls. The VB
approach is pretty confusing for many programmers.
There are a couple of potential down-sides to Delphi.
First off, the programming language for Delphi is
Object Pascal. There is nothing wrong with the Object
Pascal language, per se. Its just that people have
been moving away from Pascal over the last 5 years.
Its unclear to me whether customers are going to be
willing to change languages in order to get the benifits
of Delphi. The second potential problem for Delphi is
that it might be too late and that customers have already
made their buying decisions for application development
languages. Both VB and PowerBuilder have been on the
market for quite a while and have loyal customer bases.
Regardless of these potential problems, Delphi looks
like a great product from a technical standpoint.
Also don't look for Delphi to save the Borland company.
Wall Street expects first year sales for the product to
be in the modest $20 million range.
The keynote speech was giving by Philippe Kahn formerly
of Borland. I say former because he no longer has an
office at Borland and is instead running a new
company called Starfish. (Starfish just picked up the
old SideKick and Dashboard products for Windows.) Anyway,
Philippe gave a technical talk on performance of applications.
Apparently the show management didn't want him to talk about
the software business. I only got to hear about 10 minutes
of his speech because I had to leave due to a prior engagement.
The one little zinger that I heard from Philippe is that
he blamed the press for his downfall at Borland.
OS/2 Warp won a Jolt award in the "Special" catagory. It beat
out 13 other products including Windows NT, and Mosaic.
Another Jolt winner was Bounds Checker from Numega.
IBM was pushing OS/2 very hard in their booth. The surprising
thing is that they weren't demoing OS/2 for PowerPC in their
booth. Big mistake, IMHO. Since IBM previously demo'ed OS/2
for PowerPC at Comdex, I don't understand why they couldn't
bring it to SD '95. To see OS/2 for PowerPC, you had to go
over to the MetaWare booth who had it running along with their
native PowerPC compiler.
Another no-show at SD '95 was Taligent. Again, big
mistake. I don't think that Taligent understands very well
the software developer marketplace. They seem very inwardly
focused to me. I think that many people at the show would
be interested to see what Taligent was working on after hearing
about Tailgent for all these years.
Microsoft had no product introductions at the show. They
were showing Windows 95 in the booth along with all of
their current language products (VC, VB, OLE, Windows NT, etc.)
Motorola had a booth at the show with demos of the PowerPC.
One of the stations in the booth had Windows NT for PowerPC
running. Another station had a microscope set up and you
could look at a PowerPC die.
Intel was demonstrating a 150Mhz Pentium system in their
booth.
A bunch of the smaller tool companies who I have seen in
previous years, didn't make it to this years show. Many
of these companies specialize in DOS tools and are now
out of place in what is mostly a Windows tools show.
The VB crowd seems to be staying away too from the SD show.
I guess its because they have there own show, VBits.
Most of the Macintosh tool vendors seem to be camping
out in the Apple booth. There were maybe 7 to 8 Mac
vendors at the show. One of the Mac vendors at the show,
Metroworks, was talking about a new product that had me
scratching my head. They are going to bring out a C++
compiler for building Win32 applications. The strange
thing about the compiler is that it is a cross-compiler
that runs on a PowerMac. Why a Windows developer wants
to go out and buy a PowerMac to do compiles for Windows
applications, I can't understand. My only guess about
this product, is that they plan to do a native WIN32
version of the compiler and their IDE sometime in the
future and they want to get a small foothold in the
Windows market today.
Richard
ro...@crl.com
Smith<75070...@CompuServe.COM> writes:
>Most of the Macintosh tool vendors seem to be camping
>out in the Apple booth. There were maybe 7 to 8 Mac
>vendors at the show. One of the Mac vendors at the show,
>Metroworks, was talking about a new product that had me
>scratching my head. They are going to bring out a C++
>compiler for building Win32 applications. The strange
>thing about the compiler is that it is a cross-compiler
>that runs on a PowerMac. Why a Windows developer wants
>to go out and buy a PowerMac to do compiles for Windows
>applications, I can't understand. My only guess about
>this product, is that they plan to do a native WIN32
>version of the compiler and their IDE sometime in the
>future and they want to get a small foothold in the
>Windows market today.
>
>Richard
I think this was to slight M$. At present, you can do Mac development using
Visual C++, but you have to build on an NT machine and debug on the target Mac.
Awkward, to say the least. It's also my understanding that NT/PPC is in beta, so
having a Win32 compiler that's put out by someone other than M$ operating on a
PPC machine doesn't really seem that far out of line.
I think the reason IBM's not demoing OS/2 PPC was that (according to the
MetaWare folks), it's still a debug build...they haven't seemed to have resolved
a number of (apparently fatal) release 'features.' As a result, it runs as slow
as molasses in January (5 minutes to boot, according to the MW rep).
Roby
Marvin Lichtenthal
did not match what is on OS/2 Warp. In a few minutes I caught several
differences.
I was also surprised that MS was running Build 314 of Windows 95.
This is quite an old build? It could be significant that they did
not have anything more recent that they felt belonged in a public
demo.
--
Marvin Lichtenthal
mar...@netcom.com
Blake Stone
:
: The one new product that was introduced at the show was
: Borland's Delphi. Delphi is basically a Visual Pascal
: compiler with a built-in database. It comes in two
: flavors, a desktop version and a client/server version.
: The desktop version competes with Visual Basic and the
: client server version competes with PowerBuilder.
: << Snip >>
:
: There are a couple of potential down-sides to Delphi.
: First off, the programming language for Delphi is
: Object Pascal. There is nothing wrong with the Object
: Pascal language, per se. Its just that people have
: been moving away from Pascal over the last 5 years.
I certainly hope they haven't been moving away from Pascal in
order to move to PowerScript! Given the choice between an
object oriented variant of a moderately popular language
and a proprietary 4GL scripting language I'll take the former,
thank you.
PowerBuilder's real strength lies in the flexibility of their
DataWindow class, not the language built to support it. If
Delphi offers anything like the apparent learning curve of
PowerBuilder I would expect it to have a serious chance in the
market. (By apparent learning curve I mean "does it look feel
seductively easy to get started with" ... even a GREAT product
will die in the market if its competitors seem easier to get into
regardless of what the long term realities are).
--
-------------------------------------------------------------------------
Blake W. Stone bst...@acs.ucalgary.ca
Object Addict - Arcane Systems Ltd. 'Twas brillig, and the slithy toves
Publishers of ThreadKit Did gyre and gimble in the wabe...