[Info-vax] AXIS2/C, gSOAP

[Neil Rieck]

On Aug 20, 7:31 am, Jan-Erik Soderholm <jan-erik.soderh... at telia.com>
wrote:
> On 2010-08-20 13:10, Neil Rieck wrote:
>
>
>
> > On Aug 20, 6:46 am, Jan-Erik Soderholm<jan-erik.soderh... at telia.com>
> > wrote:
> >> On 2010-08-20 01:07, Neil Rieck wrote:
>
> >>> I've got some great news regarding gSOAP. I have written some
> >>> demonstration apps and can tell you that this thing is bloody fast.
>
> >> How are these apps runed ?
> >> USing the bultin standalone http server in gSOAP ?
> >> Using Apache or WASD ?
>
> >> Here are some values posted by Mark Daniel on the WASD
> >> maillist :
>
> >>   >  Test-bench performance is again more than competitive with
> >>   >  mod_gsoap. For 1,000 back-to-back requests of the WASD example
> >>   >  SOAP application ECHO.C/ECHOCLIENT.C ...
>
> >>   >                        Seconds  Average mS  Requests/S
> >>   >    Apache mod_gsoap      22.87        22.8          44
> >>   >       WASD gSOAPrte       5.12         5.1         195
> >>   >        WASD CGIplus       5.78         5.8         173
> >>   >            WASD CGI      46.72        46.7          21
>
> >> Are these comparable with your results ?
>
> >>> I
> >>> was a little uncomfortable with gSOAP at first and the main reason was
> >>> because it wouldn't bounce malformed XML messages. It turned out that
> >>> a relaxed mode is built in by default (so you can get a partial
> >>> solution up-and-running in a short amount of time) but you can use
> >>> option flags to put it into a few intermediate modes between relaxed
> >>> and strict. Cool.
>
> >>> Neil Rieck
> >>> Kitchener / Waterloo / Cambridge,
> >>> Ontario, Canada.
> >>>http://www3.sympatico.ca/n.rieck/
> >>>http://www3.sympatico.ca/n.rieck/docs/openvms_notes_axis2.html
>
> > I don't have the numbers with me at the moment but gSOAP (via Apache
> > mod_gsoap) was around 8-times faster than AXIS2 on Apache Tomcat.
>
> And WASD with theh persistent server ("rte") environment seems to
> be 4-5 times faster then Apache mod_gsoap... :-) From the stats
> from Mark Daniel, at least.
>
> > I
> > was running on an Alpha Server DS20e with 3 GB and two CPUs. The
> > OpenVMS flavor of gSOAP is written to link with so-called DEC
> > languages like FORTRAN and COBOL etc where apps written in those
> > programs would directly access databases.
>
> Yes, that is how we/I run. gSOAP in "server" mode.
> A C wrapper doing the gSOAP calles/APIs.
> COBOL subroutine called from the C wrapper that
> does the Rdb stuff (all application logic code).
>
> > In my case it was better for
> > me to write the whole thing in "C" then only convert SOAP messages
> > into flat files which are passed onto BASIC apps via VMS queues.
>
> "queue" is in "a real queue" or as in "mailbox" ?
>
> One could have the BASIC app written as a queue symbiont,
> I guess. But I also *guess* that you are using mailboxes, not ?

Batch queue (which we can easily stop/start whenever we want to stop
processing in-bound messages but not stop receiving them). Yeah I
know, we could use a combination of mailboxes and logical names but we
have never got around to updating this 1994 solution. (it would be a
major rewrite of a lot of stuff)

   ###

Since you brought up BASIC let ask everyone here about the future of
that language. If you play with DEC's modern FORTRAN compiler you will
notice a new compile-time switch called "/name" so you can do stuff
like "/name=as_is". A similar switch can be found in both the "C" and C
++ compilers and simplifies linking object files produced by different
languages.

I am currently using BASIC 1.6 (for Alpha) and this switch is not
available. I can see in HP's most recent road map that they have
already released BASIC 1.7 and are working on BASIC 1.8. Can anyone
tell me if the "/name" switch is available in 1.8 or might be
available in 1.8 ?

NSR