Hello
Here is the latest Caml Weekly News, for the week of February 20 to 27, 2007.
Whilst revamping our website, I took the time to improve the maze generator: http://www.ffconsultancy.com/ocaml/maze/ it is now 30% shorter and much faster, using an array instead of a functional map. It also uses continuation passing style instead of an explicit stack. But it no longer produces PostScript output. The program uses a simple depth first search. An interesting alternative algorithm involves keeping track of sets of connection cells and breaking walls to unify the sets until you have a spanning tree. That should be easy to code in OCaml but I think it will generate mazes that are much harder to solve.
I'm pleased to announce the release of PG'OCaml 0.7. PG'OCaml is a syntax extension to OCaml which allows you to issue PostgreSQL commands in a type-safe manner from within OCaml code. You can find out a bit more here: http://merjis.com/developers/pgocaml This code is released under the GNU LGPL + OCaml linking exception. Changes since 0.5: * Comprehensive support for profiling statements and analysing the results. * Gabriel de Perthuis - fixes for OMakefile support. * Don't crash if we get notice messages from the server. * More types supported. * pa_pgsql syntax can connect to multiple databases.
Xmlm is an OCaml module providing sequential XML input/output and
a persistent cursor. It aims at making non valid XML processing
robust and painless.
The sequential interface can be used to process documents without
building an in-memory representation. It also lets the programmer
translate its own data structures to an XML representation and
vice-versa.
The cursor allows to navigate and update a simple in-memory tree
representation of XML documents. Updates performed by the cursor
are persistent (non destructive).
To facilitate direct integration into projects, Xmlm is made of a
single module and distributed under a BSD license.
Project home page : http://code.google.com/p/xmlm
Your feedback is welcome,
Daniel
P.S.
Why another XML parser ?
Dissatisfaction about existing solutions either too complete and
complex or too britlle and restrictive. Besides it seems all
existing parsers force you to read the whole document in
memory. Here are some points that motivated the design of Xmlm.
1. Easy to integrate into projects without introducing external
dependencies. A single module provides everything including
documentation (ocamldoc) and the license.
2. Well documented. Features and limitations of the parser are precisely
documented.
3. Easy to use yet flexible api.
- Choice between sequential (SAX-like) or tree (DOM-like) processing.
- Construction/deconstruction of user data structures from/to xml
documents.
- Tree processing with persistent cursor (zipper).
- Simple white space handling options for character data.
- Character encodings are translated to UTF-8.
UTF-8 is the only encoding the programmer needs to handle.
- Character references and predefined entities are resolved.
Other entity references can be resolved via a user provided
callback.
- Early access to data to allow parse time data transformations.
- Parse time element pruning.
4. Robust parsing. Does not assume an xml subset.
- Supports major encodings : ASCII, UTF-8, UTF-16 (LE and BE),
ISO-8559-1.
- Parses qualified names (namespaces).
- Tail-recursive.
5. Limitations. If you need one of these things use PXP.
- Comments, processing instructions and standalone declaration are
dropped by the parser (it is a feature).
- No DTD support (but it can be extracted and written as a raw
string).
- No validity support.
Despite an apparent curse causing any server hosting it to experience catastrophic hardware failure, I have once again revived the web page for the Yaxpo XML parser: http://people.csail.mit.edu/mikelin/yaxpo Unfortunately, this should not be construed to suggest that I have any intention of updating the package; it has worked sufficiently for my purposes for the last five years. Xml-Light or the recently announced Xmlm are probably easier to use, unless you need the weird reentrancy stuff enabled by the astounding fallacy of an XML parser written entirely in CPS.
Here is a quick trick to help you read this CWN if you are viewing it using vim (version 6 or greater).
:set foldmethod=expr
:set foldexpr=getline(v:lnum)=~'^=\\{78}$'?'<1':1
zM
If you know of a better way, please let me know.
If you happen to miss a CWN, you can send me a message and I'll mail it to you, or go take a look at the archive or the RSS feed of the archives.
If you also wish to receive it every week by mail, you may subscribe online.