W3C home > Mailing lists > Public > www-dom-ts@w3.org > February 2002

Re: My stab at a distribution

From: Curt Arnold <carnold@houston.rr.com>
Date: Wed, 13 Feb 2002 08:49:05 -0600
Message-ID: <004b01c1b49d$98b57ae0$a800a8c0@CurtMicron>
To: "Dimitris Dimitriadis" <dimitris@ontologicon.com>
Cc: <www-dom-ts@w3.org>
Dimitris Dimitriadis wrote:
>general remark: the ecmascript/tests folder is very strange, contains
>files with truncated filenames (problem with file name length?), several
>similar files, eg. documentinvalidcharacterexcepti,
>documentinvalidcharacterexc.1, documentinvalidcharacterexc.2 and
>Problems during the build or file system related?

I don't see that in my local copy of the tar.gz.  Though I often see
warnings in other open source projects that .tar.gz files must be extracted
using GNU tar (or compatible) since earlier Unix tar's apparently had
filename limits.  The distribution produces both a .zip and a .tar.gz, but I
only uploaded the .tar.gz since the .zip was much larger and I didn't have
enough space on my personal home page.

>[dd] True, but we need to decide whether to have the source there, both
>for Java and ECMA, just in case people want to write their own harness.
>I'd go for including them, which is why I had them in my distribution
>sent earlier.

The Java source is provided in the dom1-core.jar.  The .js standalone code
is not usable by any known test framework.  I thought providing it would be
confusing since people may set breakpoints on the .js file, not realizing
that the running code is actually within the .html pages.

>OK, but we still don't want to release before having the patches
>built into a release of JsUnit?

I'd prefer that we don't, but this would be tolerable.

> The test matrix now has links to both the XML source and Doxygen
> documentation for each test since Mozilla doesn't display raw XML in a
> pleasing manner.
[dd] Several of the links to the Java source didn't work.

Possibly the same issue with filename truncation.

>2. index.htm on the CVS. I don't see the point for this, except if we
>manage to automate title, date creation (releases will vary as to when
>they are released) and other things to make it relevant for the
>particular distribution created

The idea is that anyone with the right tools installed can come back and
generate any specific release by checking out properly tagged sources from
the CVS and building.  In our case that depends that several external
resources remain available and unchanging (the DOM specs, JUnit, JSUnit).

I cheated (it was getting really late) and did not do a "ant clean" before
"ant dom1-core-dist" and inadvertantly provided xercesImpl.jar and
xmlParserAPIs.jar in my distribution.  The proper way to do the distribution
is to do a full checkout of the CVS sources to a clean directory and then
run "ant dom1-core-dist" to ensure that no uncommitted file is participating
in the build.
Received on Wednesday, 13 February 2002 09:49:22 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:34:04 UTC