Forum OpenACS Development: Re: NaviServer on Windows: [ns_info threads] returning unbalanced braces
one key point cold, "could", be the different TCL version.
In my installer I'm still using 8.5.18.
Soon there will be a new VS version. When that is available, I will make an update of all components.
On top of that we did not use the same compiler, the same switches.
On point (7) I probably was not clear. I will make an example, perhaps I am wrong:
on my distribution I kind of distinguish between core components and ancillary tools, e.g.
1. Naviserver, it's modules, Tcl, Xotcl, etc... all these things are considered core components, and they are built from sources using the same compiler, the same switches, the same everything....
2. ancillary binaries like Curl, Wget, Perl, and the like are not compiled together with the core components, they are kind of "external binaries".
if Naviserver calls a Tcl function and Tcl has been compiled exactly in the same way and with the same tools (and same switches, like --enable-threads, and so on...) there are for sure less problems than the ones we could have when a MS Naviserver calls a MS Tcl compiled with different switches or even worse a MINGW or a CIGWIN Tcl. What do you think?
Finally the call
works well with OpenACS and doesn't with ]po[ and I believe this is because ]po[ is more demanding in terms of timing/threading. In other words, ]po[ exposes a problem we have in Naviserver timing/threading (only on Windows).
Did you run ]po[ on your Windows test platform?