[boost] Re: [library process] should we more actively pursueuniversity thesis projects?

From: Thorsten Ottosen (nesotto_at_[hidden])
Date: 2004-11-07 06:23:05

"Valentin Samko" <boost_at_[hidden]> wrote in message
news:1902097042.20041106174726_at_digiways.com...

| TO> 4. An XML parser and generator library

I mentioned this because it is on the wish list for C++0x. Unfortunately I
don't know much about
this stuff-

| I have written a XML parser/DOM constructor while working on my PhD a
| few years ago (I needed to parse/process 100Mb-500Mb XML files as fast as
| possible), so one of the main issues was to avoid extra copies of any
| strings.

yes, there are a lot of situations in which string performance is critical. I
belive
using views into the bigger strings is the way to go.

| It can also do SAX style parsing, i.e. process incomplete
| XML documents and notify the caller about every XML tag. This is
| somewhat faster if one only needs a few tags from a huge XML document.
|
| So far, my parser is used in both, university and
| commercial environments (I distribute it under boost licence),
| so this is not a university project any more.
|
| It can be compiled by Intel C++ 7, g++ 3, VC++ 6, or higher versions.
| I have not tried any other compilers.
|
| Although it does some XML validation, this is not a proper validating
| parser, since it was more important to keep it fast and lightweight.
| It does not support XML namespaces yet, but I have plans to add that.
|
| I do not think it is ready for submission to boost yet(not enough
| comments, the code is not clean enough, and I am going to add a few
| features/change a few things next month), and I do not know whether
| anyone wants such a parser in boost.
|
| If anyone is interested, I will clean it up, add some features,
| produce documentation and prepare it for submission/review (hopefully by
Christmas).

It all sounds very nice.

Here is what I think you should do.

1. wait until the new release is out; then the list is normal again
2. in the mean time, find out about other solutions to this problem and
compare their functionality and speed and design;;
    find out if other people are working on somthing similar and see if they
want to cooperate with you (if you need more manpower)
3. make a new post explaining your work and provide liks to the code; make
sure you explain why this can become the ultimate
    xml-library
4. ask for people's oppinion and what features they want
5. then prepare it for a review submission

If you already have lost of usage experience with your code, I think it is a
great benefit; the fact that it has been used
commercially is also not bad :-)

best regards

Thorsten