[ANN] xampl 0.1.0

[Bob Hutchison]

Hi,

I've just uploaded the first release (0.1.0) of xampl for Ruby. I'm
expecting a few teething pains but I hope they'll be over with soon.

This is a ruby version of a project I've been working on in Java
since 1998 or so. The Java version has been used to help develop some
relatively large commercial software applications over the years. The
Ruby version is a much smaller body of code than the Java version,
partly due to the fact that it is new, but, really, mostly because of
the dynamic nature of Ruby.

It is in three parts. Xampl proper is in xampl-0.1.0.gem. The xampl
code generators are in xampl-generator-0.1.0.gem. And there is
documentation, examples, and unit tests in a separate download (the
documentation needs debugging too). I don't have access to a windows
machine, so I'd really appreciate it if someone could let me know
what happens there.

You can download it here <http://rubyforge.org/frs/?group_id=1034>
(and I obviously don't know how rubyforge works since only one of the
three parts is visible from the project's summary page).

Enjoy!

Cheers,
Bob

Xampl is a tool for developing Ruby programs. It facilitates the 'M'
part
of an MVC architecture. It is meant to be very easy to use,
supportive of
idiomatic Ruby usage, and mostly invisible.

Features:
- pure Ruby, no libraries required except those shipped with Ruby
1.8.2
- uses XML to declaratively describe a object structures
- uses a collection of example XML documents, no schema are
required
- supports arbitrary object graphs, not just XML-like trees
- Ruby idiomatic facilities to navigate the object structures
(e.g. named
arrays, hash, iterator support, uses blocks, etc.)
- support empty, simple (text content only), data (no or limited
text
content, but with arbitrary element content), and mixed
content (where
text and element content can be intermingled)
- maps XML namespaces to Ruby modules (with handy defaults) and
you can
suggest preferred namespace prefixes for XML output
- includes an XML pull parser (very fast)
- code generation is template driven, so the adventurous can
change what
is generated
- includes a simple but powerful and fast templating engine
- includes a simple-minded but effective hash-like class that
maintains
insertion order (fast insertion, lookup, and iteration, but at
the
expense of delete)
- naming conventions are used when generating code (but these
conventions
can be changed)
- additions to the XML will very rarely break existing code, and so
incremental development of functionality is supported
- generated code is readable and useful while debugging
- a command line tool to generate the code, or you can generate
at runtime
- XML binding to the classes (i.e. parsing an XML file will
produce an
object graph)
- can augment existing classes
- programmer adds functionality by opening the class or module, no
worries
about the code generator overwriting your changes
- object structures can generate XML output (and there is an XML
pretty
printer used for emphasising human readable XML output)
- easy to use persistence mechanism
- persistence is in terms of clusters of objects, the cluster is
updated
all at once
- lazy loading of persisted clusters
- LRU cache (of clusters)
- the changed clusters are available for inspection
- XML, Ruby, and YAML persistence mechanism (purposely no
Marshal support)
- persist to memory, or filesystem (optional persistence to an
FSDB)
- sync and rollback (even with memory persistence); this is a
weak form
of transaction support in this release
- 'automatic' and 'manual' modes of persistence ('automatic'
means that
any xampl cluster created while a persister is active will be
automatically managed by that persister, manual means that you
have
to introduce the xampl cluster to the persister)
- fast alternative to persistence that does not support rollback
- xampl objects can hold arbitrary attribute values through the
persistence 'cycle' if using Ruby or YAML persistence (even xampl
object); with XML you won't loose anything but all attributes
will
be strings after round tripping.
- multiple persisters, possibly with differing persistence
mechanisms,
are supported
- dead easy single user applications (e.g. desktop apps, CGI),
almost as
easy for multiple user applications and webapps
- you can add your own persister types (and formats for that
matter)
- general purpose Visitor pattern facility, with pre and post
visit methods,
cycle and revisit detection and handling
- lots of hooks
- many unit tests, all execute without warning when run with the -
w Ruby flag
- there are some examples, and the beginnings of a tutorial in the
form of
a series of examples that lead you through the core parts of xampl

Features that are coming soon:
- many more hooks
- more sophisticated augmentation of existing classes
- indexing and querying of the persisted object structures
- more persisters (e.g. sqlite)
- real transactions
- a better comparison function (that deals with cyclic structures
at all, and
better with non-tree acyclic structures)

Known Bugs or Potential Problems or Suspected Problems
- generated attribute names may conflict with Ruby keywords and
this is not
a nice thing to witness (this is easy to fix, but I don't want
to do this
until the next release at the earliest)
- I don't know if this will work on Windows -- I tried, but I
don't have a
windows machine to test this with -- there might be some
pathname stuff
assuming unix (but I hope not)
- some of the code is unnecessarily ugly (e.g. some of the visitor
internals)
- there are no Ruby docs (but I promise... really)
- comparison of cyclic structures will crash

License: GPL. An alternative license may be negotiated, contact me.

While the Ruby version of xampl is brand new it is based on a Java
tool that I've been developing since 1998 and in its fifth generation.
There are many features in the Java version that are not available in
this library, but that is either just a matter of time, or they will
never be needed because Ruby programs don't need the help.

The Java version of the tool has been used to write a number of large
webapps, and a few smaller desktop applications, and even an applet.

There is a Common Lisp/CLOS version coming.

 

[Robert Klemme]

Some questions just out of curiosity. No offense intended.

Bob Hutchison wrote:

Xampl is a tool for developing Ruby programs. It facilitates the 'M'
part
of an MVC architecture. It is meant to be very easy to use,
supportive of
idiomatic Ruby usage, and mostly invisible.

Features:
- pure Ruby, no libraries required except those shipped with Ruby
1.8.2
- uses XML to declaratively describe a object structures
- uses a collection of example XML documents, no schema are
required

Interestingly some days ago I thought about how J2EE is really two
languages (Java for implementation, XML for configuration) and how Ruby
can do both very neatly because of its clean syntax. From your
experience, does it really pay off to use XML here? Or is ist just
because you want to be able to read in arbitrary files and use the content
as Ruby objects?

- XML, Ruby, and YAML persistence mechanism (purposely no
Marshal support)

Why is that? What's the reasoning behind this?

Kind regards

robert

 

[Bob Hutchison]

Some questions just out of curiosity. No offense intended.

No problem. You've asked a couple of interesting questions.

Bob Hutchison wrote:



Interestingly some days ago I thought about how J2EE is really two
languages (Java for implementation, XML for configuration) and how
Ruby
can do both very neatly because of its clean syntax. From your
experience, does it really pay off to use XML here? Or is ist just
because you want to be able to read in arbitrary files and use the
content
as Ruby objects?

Yes it does make sense I think. How do you express succinctly the
relationship between a bunch of live objects? Amusingly, the hyper-
verbose XML can actually do that It all depends on how you
interpret the XML. XML can reflect a containment/composition/
aggregation relationship about as tersely as you can get in text.

Xampl is not about using XML for configuration, but for 1) describing
dynamic relationships between objects; and 2) persistence.

I agree with you that much of what XML, or YAML for that matter, is
used for can be better done in Ruby (or Python, or any other dynamic
language), certainly the way XML is used conventionally. Though xampl
changes the picture a bit.

The other thing xampl does is makes it a lot easier to deal with XML
based protocols. I've provided an example with the distribution (and
discussed in the documentation) of how to use xampl to read an Atom
1.0 or 0.3 feed over the internet, cache the feeds, extract the
entries, convert 0.3 entries to 1.0, aggregate in a new 1.0 feed
paying attention to ordering, and serve it up via WEBrick. Xampl
really helps in this kind of thing.

This is actually a really good question and I should prepare a better
answer. The background for what I'll say is documented in the
download file, but there is no direct answer to this question there.

Why is that? What's the reasoning behind this?

1) Upgrades of marshalled objects can be really nasty, using a text
form that xampl has control over while parsing, you have a chance of
intervening at an opportune time, making your life a lot easier.
2) You can read XML, Ruby and YAML.
3) You can use the data easily independently of any xampl program
(e.g. in a text editor
4) You cannot control how much of the network of reachable objects is
written to one file (this is very important in xampl because of its
clustering capabilities).

There are more reasons that I've forgotten, these are enough for me.

Cheers,
Bob

 

[Robert Klemme]

No problem. You've asked a couple of interesting questions.


Yes it does make sense I think. How do you express succinctly the
relationship between a bunch of live objects? Amusingly, the hyper-
verbose XML can actually do that It all depends on how you
interpret the XML. XML can reflect a containment/composition/
aggregation relationship about as tersely as you can get in text.

Well, you can describe those relationships in Ruby, too. In fact, nearly
every Ruby program does it by creating / assigning member variables.

Xampl is not about using XML for configuration, but for 1) describing
dynamic relationships between objects; and 2) persistence.

"describe" persistence or "implement" persistence? It seems rather the
latter.

I agree with you that much of what XML, or YAML for that matter, is
used for can be better done in Ruby (or Python, or any other dynamic
language), certainly the way XML is used conventionally. Though xampl
changes the picture a bit.

The other thing xampl does is makes it a lot easier to deal with XML
based protocols. I've provided an example with the distribution (and
discussed in the documentation) of how to use xampl to read an Atom
1.0 or 0.3 feed over the internet, cache the feeds, extract the
entries, convert 0.3 entries to 1.0, aggregate in a new 1.0 feed
paying attention to ordering, and serve it up via WEBrick. Xampl
really helps in this kind of thing.

This is actually a really good question and I should prepare a better
answer. The background for what I'll say is documented in the
download file, but there is no direct answer to this question there.

So one of the strengths of xampl is to deal with existing XML data.

1) Upgrades of marshalled objects can be really nasty, using a text
form that xampl has control over while parsing, you have a chance of
intervening at an opportune time, making your life a lot easier.
2) You can read XML, Ruby and YAML.
3) You can use the data easily independently of any xampl program
(e.g. in a text editor
4) You cannot control how much of the network of reachable objects is
written to one file (this is very important in xampl because of its
clustering capabilities).

There are more reasons that I've forgotten, these are enough for me.

Well, that's quite comprehensive! Thanks for the quick and exhaustive
answer!

Kind regards

robert

 

[Kirk Haines]

You can download it here <http://rubyforge.org/frs/?group_id=1034>
(and I obviously don't know how rubyforge works since only one of the
three parts is visible from the project's summary page).

I'd love to take a look at xampl. Could you, though, upload an archive of it
that doesn't depend on gems?


Thanks,

Kirk Haines

 

[Bob Hutchison]

Absolutely. Good idea. Can you point me to how you deal with require
statements that work with either. I think I saw something like that
but I can't find it now. How would you like it done, since you've
asked first?

Cheers,
Bob

 

[Bob Hutchison]

Well, you can describe those relationships in Ruby, too. In fact,
nearly
every Ruby program does it by creating / assigning member
variables.

Sure, but not so tersely and not so easily understood by non-
programmer. Most people are capable of recognising a document in a
slightly unfamiliar form and how to read one, and XML is something
they can recognise. You can design a xampl system by modelling paper
pushing.

That's not to say xampl won't be able to work directly on Ruby
classes at some point Though that is a ways off I think.

"describe" persistence or "implement" persistence? It seems rather
the
latter.

Implement persistence as an option.

So one of the strengths of xampl is to deal with existing XML data.

Yes, and producing XML. Though, and I really can't stress this
enough, it even helps *a lot* if XML is only used as a description
and never in practice.

 

[Austin Ziegler]

Absolutely. Good idea. Can you point me to how you deal with require
statements that work with either. I think I saw something like that
but I can't find it now. How would you like it done, since you've
asked first?

I recommend looking at the PDF::Writer Rakefile (in the CVS
repository) as I automatically create a gem, a .tar.gz, and a demo
tar.gz from the rakefile and even upload it to RubyForge.

I hope to be submitting a new version of the Rakefile that has a minor
fix for the .tar.gz creation (the .tar.gz is built properly, but gzip
reports that the compression wasn't completed; there is no damage to
the file, it's mostly a spurious warning) in the next couple of days.

-austin

 

[Kirk Haines]

Absolutely. Good idea. Can you point me to how you deal with require
statements that work with either. I think I saw something like that
but I can't find it now. How would you like it done, since you've
asked first?

I've made a couple gems, but I'm not really a gems user now, so I can't really
tell you, except to say that I would like to be able to download a tar
archive of xampl so that I can try it out without having anything that I do
with it forced to depend on gems.

I'd look at what Austin Ziegler has done with his PDF::Writer package. It's a
good practical example of how to target both gems and non-gems systems about
as cleanly as it currently possible.


Thanks,

Kirk Haines