Thanks Peter,
Your questions and comments have led me to rethink this a little.
1. For temporary purposes, I am going to create a GitHub project that carries
some specimen and exemplary materials for a startup Document Interop Assessment
project. This is an easier place to demonstrate my thinking and also not
clutter up the Corinthia repo with material that will go dead and should not
clutter up the history. (For various reasons, I would much rather have this be
in an SVN repository because of how repository-level versioning is handled
automatically and how HTTP access into SVN works well. I also have to satisfy
myself about using Markdown instead of HTML, since that creates more
dependencies on any server housing the materials. I think I'll check to see
how HTML renders on web access to a GitHub repository.)
2. The Interop Assessment material is not really something that is produced in
releases. There might be companion utilities that have releasable source code
and convenience binaries. However the Interop Assessment material could become
quite extensive and it makes no sense to have there be some sort of overall
release cadence.
I need to think what would be an appropriate place to house this that is
not tied to the release cadence of some project.
(I also think this is a matter for Corinthia itwself, in that there are
multiple components in what is a kind of suite of materials and having an
overall release of the code base might be a little peculiar. It seems to be
the wrong level for versioning of stuff.)
Perhaps it is better to think of this as a kind of library project, where a
big part of the library is a collection of data, documentation, and
instructions as well as a variety of (small?) utilities. Maybe code that is
developed for generic treatment of the material and assessment of documents,
conduct of tests, etc., is more appropriate for something like Apache Commons.
Then there are all the cases and their data.
I am going to ask around Apache about this. Maybe some other places.
- Dennis
-----Original Message-----
From: Peter Kelly [mailto:pmke...@apache.org]
Sent: Monday, June 15, 2015 21:52
To: dev@corinthia.incubator.apache.org
Subject: Re: [DISCUSS] Corinthia Home for ODF Interoperability Assessment
> On 16 Jun 2015, at 3:11 am, Dennis E. Hamilton <dennis.hamil...@acm.org>
> wrote:
>
> I want to start building specimen documents that fit the model of
> interoperability assessment that is sketched (sketchily) at
> <https://cwiki.apache.org/confluence/display/Corinthia/ODF> under the "ODF
> Conformance/Compliance Assurance helix."
>
> My thought is to create a branch having a folder, "InteropAssess" with
> subfolder "ODF" to start a subtree of folders that develop specimens that
> demonstrate particular aspects of ODF documents. These can be used as test
> suites but are not intended to be the same as ad hoc tests created to
> exercising particular Corinthia and DocFormats functions. Rather they are
> addressed to the standards and any profiling of processors, with DocFormats
> being only one.
[ ... ]
What sort of data volume do you think these are likely to consume? I’m just
thinking about repository size and keeping it manageable (to allow people to
quickly clone the repo if they want to build it). If it’s only a few Mb or so,
I’d say put them in the main repository, otherwise it might be more appropriate
to store these in a separate repository. Do you know if Infra supports multiple
repositories per project?
> MY HESITATION
>
> I don't quite like putting these in a Git repository because it is important
> and useful to cross-reference among the materials and I am not clear how that
> can happen in a non-web repository system. I do know how to make it work
> with a SubVersion repository because one can use the fact that the SVN is
> part of a web site and can be navigated with a browser. One can even put
> HTML pages in an SVN repository and use (relative) links to cross-reference
> among the material.
>
> That is an extremely valuable way to do what I have in mind.
>
> Is this possible with the Corinthia Git repository?
Yes - though with the caveat that it depends on there being a particular server
that contains a clone of the repository. For Corinthia, you can access the
files here:
https://git-wip-us.apache.org/repos/asf?p=incubator-corinthia.git
And to reference a specific file:
https://git-wip-us.apache.org/repos/asf?p=incubator-corinthia.git;a=blob_plain;f=DocFormats/CMakeLists.txt;hb=377421ecf076e553beedc075e2baef65a3e7e3b0
The main problem is that these URLs are not necessarily stable;
git-wip-us.apache.org will presumably become git-us.apache.org at some point,
and upon graduation our repository will be called corinthia instead of
incubator-corinthia.
Another options - since our website is just static files, we could
alternatively host the files there (while still storing the documents in the
repository, and having a script which copies up the files to the site. However
this still has the problem of URL stability -
http://corinthia.incubator.apache.org would become http://incubator.apache.org
upon graduation. I’m not sure how important URL stability is to you for this
particular use case; if it’s not critical then we should be ok with either of
these options.
Regarding the way in which the HTML structure on the website is done, this
would not have an impact, as the files could simply be placed in a separate
directory and be linked to from the main page. There’s nothing special or
abnormal that would need to be done here.
—
Dr Peter M. Kelly
pmke...@apache.org
PGP key: http://www.kellypmk.net/pgp-key <http://www.kellypmk.net/pgp-key>
(fingerprint 5435 6718 59F0 DD1F BFA0 5E46 2523 BAA1 44AE 2966)
