Re: [boost] [Boost Review] Property Tree Library

From: Sebastian Redl (sebastian.redl_at_[hidden])
Date: 2006-04-23 10:00:15

Tom Brinkman wrote:

>Poor points.  Getting this library thing ready by version 1.35 or any other
>version is my last concern.  I want it to be correct.  Its not acceptable
>that that this fairly large library will not get the full review that is
>necessary for a boost library.  The scope of this library is too diverse for
>it that too occur in only a 10 day review period.  It is very, very common
>for libraries to need a second or even a third review.  I think that this
>library falls in that category of libraries that will need multiple reviews.
>  
>
So let's extend the review period if the library is too large to be 
reviewed within 10 days. But splitting it up makes little sense to me.

>As this seems to be opinion that you share with one or two others, please
>educate me.  Why does the "property tree" require a parser.
>
Because, as I said, it is inappropriately named. The "property tree" 
library submitted is not just the data structure. It is, to me, a 
framework to load, store, transport and save simple hierarchical 
configuration (or similar) data. As such it is useful to me. As such I'm 
already using it. As such it needs the parsers.

>I suspect that those of you who want the parsers included with this library
>are just trying to sneek an XML parser through the review process without a
>full review.  Please tell me that I'm wrong on this point.
>  
>
You are wrong. I don't particularly care about the XML parser 
specifically. Replace it with a Java .properties parser, and I'll be 
just as happy. If ever I want an XML parser in Boost (and I want one, 
make no mistake), I don't want it some half-arsed thing that understands 
only a minor subset of XML. I want it a full-blown effort, providing 
generic pull- and push-parsers, a data structure complex enough that a 
DOM can be built on top (that could perhaps be a seperate effort, 
Boost.GenericTree), understanding namespaces and full DTD validation. 
Eventually, the library ought to evolve to have full Schema (and perhaps 
RelaxNG and Schematron) and XInclude support. In other words, an effort 
as complete as the parsers that come with Java. A parser on the level of 
PropertyTree's is so basic that I might as well replace those XML files 
with INFO or JSON.

>This library will eventually get accepted, I'm fairly sure of that, possibly
>even portions of the library will be approved this time around.  What is
>your rush to get this one through.  Lets take our time and get it right.
>  
>
I have no rush to get it through. If the whole library is ready by 1.36 
or 1.37 or whatever, that's fine with me. But I want to get it through 
as a whole.

Sebastian Redl