mckenney@talos.pm.com (Frank McKenney) (05/03/90)
The following message describes experience with the Microsoft
Programmer's WorkBench 1.0, and the associated C Compiler 6.0.
Those interested in neither can just hit "n".
Caveat: the following are initial impressions of Version 1.0 of a
complex piece of software, made by a reasonably competent user, but
one who has had only 5 days' experience with the product. The items
below might equally be results of mis-read, missing, or incorrect
documentation, product deficiencies, actual bugs, or my own
ignorance. Not to mention possible klutz errors...
Caveat 2: my purpose was to learn enough about PWB to be comfortable
using it for editing, compiling, linking, and testing. I have left
the fancier features (Source Browser) alone for now.
Equipment: 10MHz AT Clone, 5 Mb RAM, MS Bus Mouse, Video 7 VGA,
ST-296N, ST-251-1, DOS 3.3, TMP=(2Mb RAM disk)
--------------------------------------------------------------------
4/27 - Installation. DOS/Real Mode only (for now). Quick Help?
Sure!. Small Memory Model library only. Six megabytes gone. Ouch.
--------------------------------------------------------------------
4/28 - U1354 (undocumented) error from PWB. Since PWB's directory
(D:\MSC6\BIN) had been added to my PATH, I assumed that I could run
it from anywhere. When I tried it, however, I got error U1354 unless
I started it from within D:\MSC6\BIN. Neither the PWB online help
nor the QH.EXE progam could tell me what this error meant. Then I
tried my first compile, and got the same error code. Next I tried
running NMAKE outside PWB. NMAKE claimed that it couldn't find
CL.EXE or LINK.EXE, although I could execute them from the DOS
command line without problem.
The removal of a TSR utility named DPATH finally fixed the problem.
Apparently DPATH's assistance was confusing the EXEC method used by
the new utilities.
--------------------------------------------------------------------
4/29 - Startup and exit timings.
PWB takes about 30 seconds to come up, and 3-5 seconds to shut down
(not counting prompters). Shelling out to DOS ("Run DOS Command"
COMMAND) takes about 6 seconds, and it takes about 24 seconds to
return (timings done with 3 open files totalling about 22K).
Using "File/DOS Shell" is a few seconds faster, and returns into the
PWB environment about 5 seonds faster.
--------------------------------------------------------------------
4/29 - TMP vs. TEMP oddity. I have an environment variable named TMP
set to point to my RAM drive (VDISK.SYS), but no TEMP. PWB seems to
be working just fine, and various messages indicate that the
contents of TMP are being used.
--------------------------------------------------------------------
4/30 - I really like the CodeView VGA 50-line mode. CV now has four
adjustable windows (Watch, Source, Memory, and Command), with the
result that there is too little room on a 25-line display for a
reasonable number of Watch statements.
--------------------------------------------------------------------
4/30 - I seem to occasionally wind up with an "open" help window in
my "Files" list, and can't figure out how to get rid of it. It does
disappear when I shut down PWB and bring it back up again.
--------------------------------------------------------------------
4/30 - Trying to set up to run in Debug mode can be tricky. There
are several different places where switches have to be set, and it
took a while to find all of them. Unfortunately, the only way of
checking was to re-compile and go into CodeView, and then be told
that there was no symbolic information because it was a "packed"
file. A "master switch" would be nice.
--------------------------------------------------------------------
4/30 - PWB does a nice job of keeping track of all of my open
program files. I only wish it also kept my MAKE file - I have to
respecify <that> with "Make/Set Program List" each time I bring PWB
up. This seems inconsistent.
--------------------------------------------------------------------
4/30 - I keep getting (intermittently) "C1069: write error on file
xxxx.obj" when compiling. Other programs like this drive just fine,
and a recompile always succeeds. To date, this has always occurred
with other compile errors.
--------------------------------------------------------------------
5/1 - There appears to be a problem with the "incremental compile"
feature. If I have an error, fix it, and then recompile, I occasion-
ally get the message "C4140: 'function' redefined, preceeding
references may be invalid" for each function which follows the
function fixed. A "Make/Rebuild All" fixes the problem.
--------------------------------------------------------------------
5/1 - Time to "Open" files seems reasonable. Switching between
already "open" files goes fairly quickly. Using the vertical
"slider" to move through the active file is very quick. Switching
<windows>, on the other hand, can take as long as 10 seconds.
--------------------------------------------------------------------
5/2 - If I leave PWB (either exiting, or to CodeView) after making
changes to multiple files, I will get a prompter labelled "File has
changed!" for the current file. If I respond "Y" to save it, I will
then get <another> prompter asking if I want to save "All Files".
Since I generally want to save everything, the ordering of these
prompters seems backwards.
--------------------------------------------------------------------
5/2 - Shelling out to DOS. PWB appears to write most of itself out
to files in the directory SET by TMP:
PWB08090 V00 163848 5-02-90 1:01p
PWB SHL 122 5-02-90 1:01p
leaving only 14K resident when a DOS command is "Run" from the menu.
This is fairly impressive. Using the "File/DOS Shell" method creates
an additional file, but uses 4K less (10K resident). The "Run/Run
DOS Command" method does <not> retain the last command used. Sigh.
--------------------------------------------------------------------
5/2 - Cursor, bell: minor irritations. During file load and other
lengthy (>.5 sec) operations, the cursor tends to jump around on the
screen, rather than being blanked. Also, there is a "beep" that I
<assume> signals the end of a rebuild, but it occurs 10-15 seconds
before I get any other results. Sounds like a left-hand/right-hand
problem.
--------------------------------------------------------------------
5/2 - Mouse positioning. It may take several clicks for the mouse
to reposition the text cursor. This seems to occur mainly (only?) at
the left margin. I suspect that the PWB mouse support is giving me
the "accurate" response ("You can't click on a window border")
rather than the "useful" one ("Oh. You're <actually> clicking on the
window border character, but since that isn't useful, I'll <act as
though> you were at the left margin"). The problem, of course, is
that the window border is a thin vertical line, so it's hard to tell
the difference between the blank portion of the vertical line's
"character space" and a blank character to its immediate right.
Quick fix: I used "Options/Editor Settings" to set the window border
from 17 (white on blue) to 74 (red on white), making the "real"
border more obvious.
--------------------------------------------------------------------
5/2 - PWB crashes on "run-time error R6003 - integer divide by 0"
I loaded a copy of this document (MSC6.DOC, in XyWrite III format)
into PWB to make a few minor changes. It was a little tricky, since
my "paragraphs" exceed PWB's "line" length limit, but it looked OK
on the screen. When I tried to re-save the result, I got a prompter
with the messages "Cannot rename FILE.$!" and "No such file or
directory". After several more prompters, PWB finally terminated
with an R6003 run-time error. Fortunately, MSC6.BAK (the XyWrite
backup) and MSC6.$ (the PWB backup) were both still present. Guess:
PWB doesn't gracefully handle "long" text lines.
--------------------------------------------------------------------
Overall reaction: favorable. I believe that Microsoft has defined
a framework which will last for a while (i.e. four or five years,
a long time in this business). PWB does appear to have a few bugs,
it doesn't run as fast as I would like for some things, and it makes
me wonder when I'll wind up buying a 200Mb disk, but I <can> use it,
as shipped, to productively develop software. And that's really what
I need. Thanks, Microsoft.