NCSA Mosaic v2.0alpha3 (originally in alt.winsock)
Jon E. Mittelhauser
>Appears flawless to us (just a few units are now on the new ver 3-4). GPFs are
>not occurring in our environment! While imperfect (doesn't print gifs plus
>text as they appear on a page), it sure is nice to be able to print the text.
>I suggest that GPFs on the previous ver are also uncommon (we have 6-8 people
>who use it regularly). We are running 486/50s, a couple of 486/66s, a lone
>pentium, 8-16 mb RAM, units are from Zeos and Gateway, SMC Elites and 3C509s,
>using novell and Trumpet Winsock. This represents a pretty widely varied
>environment... and we are quite stable. You are almost certainly experiencing
>an underlying hardware and/or software prob.
Thank you. I would, however, note that you are running on nice machines with
modern hardware (esp. ethernet) - similar to that on which it is developed..
I feel the need to defend Mosaic. The fact is (and you don't want to hear
this), most of the problems which are being reported are not Mosaic's fault.
I am referring to the people who claim constant GPF's and the inability
to connect to anything (or only once or twice)...
The difficulty is that there are about a million independant variables
coming into effect here and it is almost impossible to track everything
down. Mosaic *is* more complex than most any other network software which
is running on your machine and it reveals problems that are present. The
major problem is that Mosaic is not at all graceful about reporting errors
and is not _at all_ graceful about recovering from problems.
The reason for that is simple. It is developed by only two people (one of
whom is a student) and on two identical - correctly configured machines.
I would like to offer the observation that people with modern computers
and especially ethernet cards (eg. 3c509's) seem to have a lot less problem
than older machines and 3c503's...The fact is, all Mosaic does is call
functions in the Winsock. It is entirely dependant on this being 100%
correct. Some interesting observations that I have noted from people who
were having problems and figured them out:
1) The TCP window size was too large/small and Mosaic was not as
graceful as MS_FTP at recovering from bad packets
2) The network memory was not excluded in the memory manager line
of the HIMEM/QEMM/MEMMAXER config.sys line. I think that this
could be a huge one! An advanced card like 3c509 auto-configures
itself but older cards may be misconfigured/ not excluded.
3) People have reprted that chanign video mode/video driver caused
all their problems to vanish...poof
The fact is that there are thousands of people who use Mosaic on almost
every conceivable configuration around. Mosaic is picky. Much more
so than if it had been commercially developed. If you are having major
problems, please do not assume that it cannot be your configuration
because X other program works. We are dealing with the complex interactions
between a program, winsock, network card, video card, and memory...
This is _not_ to imply that Mosaic is perfect... :^) Besides the bugs
which were known before this release, I have seen at least three major
reports which are obviously Mosaic's problem:
1) turning underlining off
2) can't resize when no document loaded
3) some problem with the TEMP file path...
Plus the obvious screen and printing problems. The fact is, however, that
we have one type of video card and one printer to work with...not a very
good test suite...This was the first 32bit version and _is_ still labeled
an ALPHA version.
As for the slam that we should have fixed bugs rather than move to 32bit,
I will refrain myself. Windows is moving to 32bit. NT and Chicago are
32 bit. The 32bit version is at least 3x as fast decoding inlined
images. The 32bit version will be much easier to progarm in the future.
Moving to 32bit fixed a number of bugs which couldn't even be theoretically
fixed in the 16bit one (e.g Mosaic 2.0a2 could not handle *very* large
documents because the scrollbar functions used an integer to mark the
position. A very large document was more than 2^16 pixels long and the
scrollbar would freak. In win32s, an int is 32bits like a real OS and
this problem vanished...).
As for speculation w/ regards to Spry, et.al: NCSA is not giving anyone
an exclusive licence. There _will_ be commercial versions of Mosaic.
Probably from multiple sources. There will (also) be a free version
from NCSA. In addition, I expect that there will be another round of
WWW browsers coming out which are independant of NCSA and Mosaic. The
appeal is clear and many things can be done better than they were with
any of the versions of Mosaic (w/ sufficient resources!). This is very
new technology. I expect by the end of the year that it will be much
more straightforward to get TCP/IP on a desktop and to get WWW software
running without any problems...NCSA never expected to get into the
position of writing software that hundreds of thousands of people would
be using (and complaining about... :^). The whole mentaility is as a
small research lab- not a major software company. Have patience. Offer
suggestions- not flames. Make sure that _your_ setup is happy and try
tweaking different parameters to make things better.
-Jon