I am writing because i have seen many words spent about a "boost engine" as a way to promote boost. I am an average programmer involved in real life consulting for some big bank firms in Italy and i would like to share my humble point of view. One of the main markets that drives the success of a language is that of enterprise applications. Today the market of enterprise application has moved from desktop and client-server systems to three-tier web applications , so the surge of garbage collected languages like java and c#. In my opinion the success of these languages is strongly depends on the fact that they provide a standard programming environment with a standard library which is very rich in functionality. As today C++ miss a freely available library that compares to the library of java or .NET. True you can find libraries for almost every kind of needs one has, but what is missing is a freely available, complete and coherent library with all the functionality one can find in these languages. Boost success is also due to the fact that it sets a standard for C++, that it contains a coherent sets of libraries and that it covers some of what is already available in other languages. Anyway i see that still boost miss some pieces that are very important for enterprise applications: for example a logging library and a database library. A synchronous networking library could also be needed by some applications. I believe that any effort to promote boost could be better spent on developing this kind of enteprise code. Regards, Luca
luca regini wrote: [...] Yes, more application level libraries are important.
Anyway i see that still boost miss some pieces that are very important for enterprise applications:
for example a logging library
Yes, and there have been various attempts to create a logging library. But the price to pay for coherency, to the C++ standard libraries in the case of Boost, is that it takes time and many times multiple tries to come up with such libraries.
and a database library.
Since this is a particular interest to me I must ask... What do you consider as a "database library"? I ask because there are numerous types of databases, and numerous interfacing to "data".
A synchronous networking library could also be needed by some applications.
Could you be more specific as to what you mean by "synchronous"? And how is that not covered by the ASIO library?
I believe that any effort to promote boost could be better spent on developing this kind of enteprise code.
They don't seem to be exclusive activities to me. Do you have other libraries in mind than the above? It would good to know what kinds of libraries *users* want. And more specifically to add this information to the wiki [**], so that when volunteers are looking to do work for Boost they can have a ready set of ideas to work on. Specifically this is something we really need if we want to continue participating in the Google Summer of Code program. [**] There's an existing list of library requests in the wiki <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?WantedLibraries>. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo
Rene Rivera wrote:
luca regini wrote: [...]
Yes, more application level libraries are important.
Anyway i see that still boost miss some pieces that are very important for enterprise applications:
for example a logging library
Yes, and there have been various attempts to create a logging library. But the price to pay for coherency, to the C++ standard libraries in the case of Boost, is that it takes time and many times multiple tries to come up with such libraries.
I hope, in time we will be able to make a proposal in this area.
[**] There's an existing list of library requests in the wiki <http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?WantedLibraries>.
Is this page moderated or anyone (me, for example) may add his request there? Anyway, it might be better if some kind of link ("Missing something in Boost? Let us know!") was on the www.boost.org main page. The link could lead to Trac page in order to collect tickets for feature requests.
luca regini wrote:
Anyway i see that still boost miss some pieces that are very important for enterprise applications: for example a logging library and a database library.
SOCI was proposed to be boostified. I don't know what's the status of this.
A synchronous networking library could also be needed by some applications.
The fact that asio does asynchronous I/O doesn't mean that it can't do synchronous I/O too.
participants (4)
-
Andrey Semashev -
luca regini -
Mathias Gaunard -
Rene Rivera