You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cocoon.apache.org by Ulrich Mayring <ul...@denic.de> on 2000/10/16 13:04:05 UTC
Re: [C1] contrib directory
Robin Green wrote:
>
> (1) - Put everything in the core distribution. But I don't want to have to
> test it all, and it would be bad to include any old junk without testing it
> first.
I can only test my logicsheets, if I know what to test against. I've
said this a couple of times before: cocoon needs reference platforms it
is tested against.
> (2) - Put them in a contrib directory in CVS but make it clear they _may_
> be
> flaky; move them into the main distro if and when they are reasonably
> "proven".
Why should I put something in a contrib directory, when I could just as
well test it against the reference platforms and put it in the distro?
On the other hand if we have no reference platforms to test against,
what is the difference between contrib code and distro code? Both are
untested. So why do we need a contrib directory at all?
> (3) - Links only in the documentation (this would have to be done anyway in
> cases where contributors weren't willing to "sign" copyright away to the
> ASF
> and release it under the cocoon license, I think). Again make it clear that
> "these are not maintained by the Cocoon project and we offer no guarantees"
> etc. I'm happy to do this right away if (1) or (2) are not agreed upon.
Code only makes sense, if it is used. Stuff in the main distro is more
likely to be used than stuff linked to from documentation that no-one
reads :-)
But I don't mind if you link to my taglibs from the docs.
Ulrich
--
Ulrich Mayring
DENIC eG, Systementwicklung
Re: [C1] contrib directory
Posted by Ulrich Mayring <ul...@denic.de>.
Paul Russell wrote:
>
> And even more funky would be if you just e-mailed the jars to
> a test address which distributed them, tested them, returned
> the results. Ambitious? Me?
I think the test server should also automatically write docs, if it
works :-)
Ulrich
--
Ulrich Mayring
DENIC eG, Systementwicklung
Re: [C1] contrib directory
Posted by Paul Russell <pa...@luminas.co.uk>.
On Tue, Oct 17, 2000 at 12:42:35AM +0200, Uli Mayring wrote:
> On Tue, 17 Oct 2000, Uli Mayring wrote:
> > I don't know what an XP style test environment is, but making standard
> > testfiles sounds like a great idea.
> Come to think of it: even better would be if the Apache group or other
> volunteers set up reference systems somewhere on the Internet and everyone
> could use them for testing. I'd submit my taglib in some way to a machine
> running some exotic configuration and would see right away if it works.
And even more funky would be if you just e-mailed the jars to
a test address which distributed them, tested them, returned
the results. Ambitious? Me?
Paul
--
Paul Russell <pa...@luminas.co.uk>
Technical Director, http://www.luminas.co.uk
Luminas Ltd.
Re: [C1] contrib directory
Posted by Uli Mayring <ul...@denic.de>.
On Tue, 17 Oct 2000, Uli Mayring wrote:
> I don't know what an XP style test environment is, but making standard
> testfiles sounds like a great idea.
Come to think of it: even better would be if the Apache group or other
volunteers set up reference systems somewhere on the Internet and everyone
could use them for testing. I'd submit my taglib in some way to a machine
running some exotic configuration and would see right away if it works.
Ulrich
--
Ulrich Mayring
DENIC eG, Softwareentwicklung
Re: [C1] contrib directory
Posted by Uli Mayring <ul...@denic.de>.
On Mon, 16 Oct 2000, Paul Russell wrote:
> Is there any way we can set up an XP style test environemnt?
> If logic sheet writers can include a set of test documents for
> their taglibs that are locatable by the test suite, then the
> above becomes much more trivial - people just hand the test
> suite the jar and let it get on with it...
I don't know what an XP style test environment is, but making standard
testfiles sounds like a great idea.
Ulrich
--
Ulrich Mayring
DENIC eG, Softwareentwicklung
Re: [C1] contrib directory
Posted by Paul Russell <pa...@luminas.co.uk>.
On Mon, Oct 16, 2000 at 02:54:05PM -0400, Claude Warren wrote:
> Perhaps the answer to the testing on reference platforms is to have a list of
> volunteers that run the specified platforms. With the installation procedures
> in hand, the software on disk, and xml page(s) that exercise the logicsheet,
> it should not take long test the logicsheet in question.
Is there any way we can set up an XP style test environemnt?
If logic sheet writers can include a set of test documents for
their taglibs that are locatable by the test suite, then the
above becomes much more trivial - people just hand the test
suite the jar and let it get on with it...
Paul
--
Paul Russell <pa...@luminas.co.uk>
Technical Director, http://www.luminas.co.uk
Luminas Ltd.
Re: [C1] contrib directory
Posted by Claude Warren <cl...@histio.org>.
Perhaps the answer to the testing on reference platforms is to have a list of
volunteers that run the specified platforms. With the installation procedures
in hand, the software on disk, and xml page(s) that exercise the logicsheet,
it should not take long test the logicsheet in question.
Ulrich Mayring wrote:
> Ross Burton wrote:
> >
> > I'll make a stab at a starting point for the reference platforms:
> >
> > 1) Windows 98, Tomcat 4 (standalone), Sun's JDK 1.2 reference
> > implementation.
> > 2) Redhat Linux 6.2, Tomcat 4 (standalone), Sun's JDK 1.2 (aka
> > Blackdown)
> > 3) Windows NT 4, IIS 4, whatever servet runner most people use?
> > 4) MacOS? I have no idea what systems people run here, but I know Cocoon
> > has been used on a Mac.
> > 5) BSD? Which BSD has a native Java port?
>
> I think we should not specify an OS. The reference platforms only make
> sense, when everyone has access to these platforms and can test his
> stuff. But not everyone has access to all those OSes.
>
> > Really another servlet container should be used. However, I've only used
> > Tomcat so can't comment here. Resin? JRun?
>
> In my mind definitely JServ, it has been the default servlet engine for
> cocoon before Tomcat came along. It is not a good idea to include
> commercial servlet engines, because they again are not accessible to
> everyone.
>
> > To pass the test, Cocoon must install and run with minimal fiddling by
> > following the manual, and all of the samples which will work out of the
> > box (i.e. not SQL) must run first time.
>
> Agreed. Also we should have a couple of "stub" documents for authors to
> fill out, for example install.xml or docs.xml or such.
>
> Ulrich
>
> --
> Ulrich Mayring
> DENIC eG, Systementwicklung
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cocoon-users-unsubscribe@xml.apache.org
> For additional commands, e-mail: cocoon-users-help@xml.apache.org
Re: [C1] contrib directory
Posted by Ross Burton <ro...@lineone.net>.
On Mon, 16 Oct 2000, Ulrich Mayring wrote:
> I think we should not specify an OS. The reference platforms only make
> sense, when everyone has access to these platforms and can test his
> stuff. But not everyone has access to all those OSes.
True.
> > Really another servlet container should be used. However, I've only used
> > Tomcat so can't comment here. Resin? JRun?
>
> In my mind definitely JServ, it has been the default servlet engine for
> cocoon before Tomcat came along. It is not a good idea to include
> commercial servlet engines, because they again are not accessible to
> everyone.
Sorry - I've been thinking about Cocoon 2 mainly here, which
requires Servlet 2.2. For Cocoon 1, of course JServ is the prime
container.
Ross
Re: [C1] contrib directory
Posted by Ulrich Mayring <ul...@denic.de>.
Ross Burton wrote:
>
> I'll make a stab at a starting point for the reference platforms:
>
> 1) Windows 98, Tomcat 4 (standalone), Sun's JDK 1.2 reference
> implementation.
> 2) Redhat Linux 6.2, Tomcat 4 (standalone), Sun's JDK 1.2 (aka
> Blackdown)
> 3) Windows NT 4, IIS 4, whatever servet runner most people use?
> 4) MacOS? I have no idea what systems people run here, but I know Cocoon
> has been used on a Mac.
> 5) BSD? Which BSD has a native Java port?
I think we should not specify an OS. The reference platforms only make
sense, when everyone has access to these platforms and can test his
stuff. But not everyone has access to all those OSes.
> Really another servlet container should be used. However, I've only used
> Tomcat so can't comment here. Resin? JRun?
In my mind definitely JServ, it has been the default servlet engine for
cocoon before Tomcat came along. It is not a good idea to include
commercial servlet engines, because they again are not accessible to
everyone.
> To pass the test, Cocoon must install and run with minimal fiddling by
> following the manual, and all of the samples which will work out of the
> box (i.e. not SQL) must run first time.
Agreed. Also we should have a couple of "stub" documents for authors to
fill out, for example install.xml or docs.xml or such.
Ulrich
--
Ulrich Mayring
DENIC eG, Systementwicklung
Re: [C1] contrib directory
Posted by Ross Burton <ro...@mail.com>.
Ulrich Mayring wrote:
>
> Robin Green wrote:
> >
> > (1) - Put everything in the core distribution. But I don't want to have to
> > test it all, and it would be bad to include any old junk without testing it
> > first.
>
> I can only test my logicsheets, if I know what to test against. I've
> said this a couple of times before: cocoon needs reference platforms it
> is tested against.
I'll make a stab at a starting point for the reference platforms:
1) Windows 98, Tomcat 4 (standalone), Sun's JDK 1.2 reference
implementation.
2) Redhat Linux 6.2, Tomcat 4 (standalone), Sun's JDK 1.2 (aka
Blackdown)
3) Windows NT 4, IIS 4, whatever servet runner most people use?
4) MacOS? I have no idea what systems people run here, but I know Cocoon
has been used on a Mac.
5) BSD? Which BSD has a native Java port?
Really another servlet container should be used. However, I've only used
Tomcat so can't comment here. Resin? JRun?
To pass the test, Cocoon must install and run with minimal fiddling by
following the manual, and all of the samples which will work out of the
box (i.e. not SQL) must run first time.
Now let's elaborate on this starting point until we're done!
Ross Burton