[omniORB] Borland port of 2.7.1: some good(?) news with BCB 4.0

[Mikhail Soukhanov]

Hello!

At last I have got some more time to work on Borland port of omniORB
2.7.1. I seem to have solved that thread termination/join problem. At
least with BCB 3.0, _endthread() not only terminated the thread but also
disposed of its handle -- so there was nothing left to join on. With BCB
4.0, M$VC-style _endthreadex() was added which provides the correct
semantics.

Now, all the `omnithread` examples run successfully (including
`thrspecdata`, which always failed with BCB 3.0), also all three `echo`
examples do well too, and all the utilities.

Additionally, I built the DII/DSI examples. `echo_diiclt` runs
successfully against `eg3_impl`. `echo_dsiimpl` runs successfully with
`eg3_clt` -- but only once! The second invocation of `eg3_clt` (or the
very first invocation of `echo_diiclt`) against the already running
`echo_dsiimpl` crashes the latter.

Also, `omniNames` crashes from the start (in the above examples I simply
used the previous successful version of `omniNames` from 2.7.0). The
problem seems to deal with heavy heap-allocation memory corruption which
originates somewhere well before it becomes visible to the debugger.
Though I am not well experienced in such, it looks to me very much like
some static memory was freed(?).

That is my progress by now. I will spent additional time on it early
next week. Also, I will then archive the new changes and make them
available at <http://astra.relline.ru/devel/omniORB_2.7.1-bcb_4.0.zip>
for those interested (not ready till Monday).

If anybody can provide some help or comments, I will be grateful.

Best regards,
		Mikhail
-- 
Mikhail Soukhanov <mailto:mik@astra.relline.ru.>

Laboratory of Geoinformatics, VNIIgeosystem
Warszawskoje chaussee 8, Moscow M-105, 113105 Russia
Tel.: +7(095) 954-00-31, fax.: +7(095) 958-35-22
W.W.W.: <http://astra.relline.ru./>