we don't have an Apache Tcl TLP mailing list, so I'm sending this
message on rivet-dev as this is the only development mailing list left
while the development about the code in trunk continues at slow but
constant pace I would like to collect suggestions and ideas about a
chart of possible projects to start within Apache Tcl or within Apache
Rivet itself. The idea is to go out to the Tcl'ers community and see if
some of the proposal can attract new contributors.
You don't need to vow your engagement for a project in order to propose
it, just throw in the discussion your wish list of missing tools.
here is my list
- Supporting Apache webserver on Windows systems. Actually I don't
know lately what is the real interest around the Apache HTTP web server
on Windows, but I had been asked by some Tcl'ers years ago and their
interest might haven't waned completely. Basically we need someone
fluent enough with Tcl and Unix tools ported to Windows or (as George
Petasis did years ago) develop a CMake build system that could integrate
well with the current build scripts
- We recently discussed the issue of the latency when starting up many
child worker threads/processes loading large datasets or lots of
packages. A Tcl interface to a shared memory database could be helpful
to minimize the cost of loading in memory data at start up. There are a
few in-memory databases available. For example in the ASF incubator
currently they have Apache Geode (http://geode.incubator.apache.org/)
but it will be offered with the Java binding, wheras the C++ binding
will be kept proprietary. I know Mysql has an engine to store tables in
RAM. I was thinking of something lighter, ideally light and self contained
- A few years ago I developed a framework for web development.
Initially I wanted the components of the framework to follow a certain
collaboration pattern and be fully reusable, but it proved to be an
utopian view. This framework was developed trying to make it as much as
possible object-oriented, has a short inheritance tree and the ambition
to have it made of fully replaceable cooperating classes. One of the key
feature of this pattern is to load packages and data as they are
demanded. Other features attempted to let it cooperate well with
mod_rewrite and also to be able to produce multilingual output. The code
resides on Sourceforge, but we may fork it into Apache Tcl and redesign
it. If someone is interested we may start a separate discussion on this.
(Yes, the Apache Tcl and Apache Rivet web sites were generated using
this tool.)
- Better documentation system. This is something less cool, still
having a handy and easy to manage tool to manage our manual would make
the difference. Docbook is heavy and not easy to handle and customize as
soon as you move from the trivial things.
- Mime protocol message handling: we are still relying on the original
code written by Damon Courtney. Back in 2007 some attempts were done by
Valery Masiutsin to let Rivet use Apache libapreq2 but the were soon
abandoned. The old saying goes "if it's not broken don't fix it" but I
think the current code at least doesn't handle nested attachments and
other cases; libapreq2 should be the standard in this sort of things and
we should rely on it.
...
-- Massimo
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
