Re[2]: win95 binaries

[Andrey Taranov <andrey@xxxxxxxxx>]

At Thursday, September 28, 2000, 9:50:06 AM hoenicka_markus wrote:

h> It took me a while to find out, but after a bit of thorough testing I'm quite
h> sure that this native Win32 version (from www.sscd.de/openjade/) is buggy.

h> The problem is not in the transformation, that seems to work ok. The
h> problem is rather in some cryptic error messages. onsgmls will (almost?)
h> always print "(Invalid Message)" instead of the real error message.
h> OpenJade will at times produce error messages which do not reflect the
h> actual problem. If you need to debug a non-valid document, this is not
h> exactly helpful. If you work with documents that are valid anyway (e.g. if
h> you create them with PSGML, which increases the odds enormously to get a
h> valid document), you will not see this problem and the output is flawless.
h> But you can simply try yourself. Change a start or end tag of a document to
h> some nonsense value and see what onsgmls and OpenJade report.

Sometime this year I did build OpenJade several times under NT. That is both
from CVS and OpenJade 1.3 source distribution. The problem you describe looks
very familiar to me. Such things happen when there is a version mismatch of
openjade.exe and osp134.dll or, maybe, due to some misconfiguration of MSVC
build process. I can't remember now, which cause is the real one. One more dim
reminiscence: it could be that I ran OpenJade with separately built OpenSP dlls,
yeah, that was the problem.

h> I contacted the maintainer of the native Win32 binaries. Unfortunately his
h> workload does not permit him to fix the binaries. So unless someone else
h> revvs up his MSVC (I don't have the $$), this will not be fixed anytime soon.

Well, I have some binaries available. They are OpenJade 1.3pre1 and OpenSP 1.3.4
and work fine. Built from CVS branch jade_1_3_branch on May 31st 2000. I plan to
make them available to the public, if anyone is interested.

There is a pitfall, though. These binaries were built from patched sources. This
tiny little patch fixes a bug in the RTF backend concerning some extra generated
spaces. However, in our team, we use these binaries in a production environment
for 4 months now and no problems have arisen. The description of the patch I one
send to Matthias Clasen, but got no reply. I will redirect the patch to the list
now.

-- 
Best regards,
 Andrey                            mailto:andrey@xxxxxxxxx



 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist