Re: change to test suite overview page - request for review and comments

  I've always assumed that some bright spark decided it would be a good 
idea to name the repository after the date on which it was created, and 
another bright spark decided it would be a good idea to name the test 
suite release after the date on which *it* was created, and neither 
bright spark thought about how silly the resulting combination would look.

Michael Kay
Saxonica


On 13/10/2010 7:15 PM, C. M. Sperberg-McQueen wrote:
> On Oct 13, 2010, at 10:26 AM, Michael Kay wrote:
>
>> A couple of typos:
>>
>> "tests which have not changes since"
>>
>> "considerably extending there 2002 contributions."
> Thanks; corrected these.
>
>> There is information about read-only access to the public repository here:
>>
>> http://dev.w3.org.cvsweb.info/
>>
>> No idea whether it works!
> I've added both a pointer and a brief description of one form the
> commands might take.
>
>> It perhaps should be emphasized more clearly that you need to download the last general-release ZIP file, and then download from CVS on top of that to get the changes - it's a rather unorthodox and unfriendly mechanism.
> Thanks; I've tried to say this more clearly, and I have taken the liberty of quoting
> your description, which seems a graceful way of warning the reader that the
> XSTS is set up in the way that differs from what at least many readers will
> expect.
>
> Another question arises when I walk through the steps described,
> trying to expand the description to provide a little more guidance.
>
> The table of releases includes a link to a file named xsts-2007-06-20.tar.gz;
> which makes me think that it is intended to reflect the status of the test
> suite on 20 June 2007.  (When I check the tar.gz file out of CVS, it gets
> a file system date of 6 July 2007, but the log shows it was indeed initially
> checked in on 20 June 2007.)  But when I unzip the file, what I get is a
> directory bearing the date 2006-11-06.  This may be related to the fact
> that the tar.gz file is in a directory called xmlschema2006-11-06 on the
> main W3C server, and to the fact that on dev.w3.org, the current version
> of the test suite appears to live in a directory called xmlschema2006-11-06,
> which itself is a child of a directory named 2004-01-14.  (It has siblings
> named xmlschema2002-01-16/, xmlschema2004-01-14/, and
> XMLSchemaTests/.)
>
> Perhaps my brain power is not up to the standard expected of users of
> the test suite, but I am finding the parent/child relations here a little
> confusing.   I can think of several possibilities:
>
> 1 There is a systematic design for the organization and naming of
> these directories, there is someone in the world who understands it,
> and it's important (or at least worthwhile) to carry on in the same way.
>
> In that case, someone who understands what is going on here
> needs to explain it to me, so that in the course of managing the
> test suite I can carry on in the same way and not break things.
>
> 2 There is, or was, a systematic design for the organization and naming,
> but it's not important to maintain it.
>
> In that case, it would be nice if someone could explain the
> organization and naming to me, and if we could get the explanation
> into a document for future users of this material.   But for ongoing
> work, I suspect that it would simplify life to retire the existing
> directories and make a new home for the test suite, for example as
> a top-level module in dev.w3.org, with a directory structure that
> would be documented from the outset.  That new home could have a
> pointer to the old location, but would not be responsible for
> explaining it.
>
> 3 There is not a systematic design or explanation for the naming;
> they should be regarded as opaque identifiers.
>
> In that case, I definitely would like to move the test suite to a new
> location and give it names that make some sense.
>
> Comments welcome.
>
>
>
>
>
>
>>
>> On 13/10/2010 5:04 PM, C. M. Sperberg-McQueen wrote:
>>> In an attempt to make the XSD test suite a little easier to acquire
>>> and use, for people who don't already have it and aren't already
>>> using it, I've added a section to the test suite overview page at
>>>
>>>    http://www.w3.org/XML/2004/xml-schema-test-suite/index.html#getit
>>>
>>> with instructions for obtaining a copy of the test suite.  Essentially
>>> it consists of a paraphrase of Henry Thompson's account of the
>>> expected usage pattern for the test suite, in his message of 7 June
>>> to this list:
>>>
>>>   http://lists.w3.org/Archives/Public/public-xml-schema-testsuite/2010Jun/0005.html
>>>
>>> Those who understand how things are supposed to work in their
>>> current configuration are asked to review the new section for
>>> factual accuracy; those who don't are asked to review it for
>>> clarity (and optionally to try the process and see if it works for
>>> them as they would expect it to).
>>>
>>> One question that arises immediately for me, as I try to execute
>>> the steps:  I'm asked for my SSH passphrase.  That's fine for me,
>>> because I do have an SSH key on the relevant machine; what must
>>> users do who wish to check out a read-only copy of the current
>>> material and update it?  Are they required to get an SSH key and
>>> have it installed, or is there a different way to invoke CVS so as
>>> to get the data in the CVS repository without having an SSH key?
>>>
>>> Different, that is, from the command I gave, which was
>>>
>>>     cvs -d dev.w3.org:/sources/public \\
>>>        checkout -d xsts1.1 \\
>>>        XML/xml-schema-test-suite/2004-01-14/xmlschema2006-11-06
>>>
>>> Thanks,
>>>
>>> Michael Sperberg-McQueen
>>>

Received on Wednesday, 13 October 2010 19:06:46 UTC