On 10/24/01 6:27 PM, "Sam Ruby" <[EMAIL PROTECTED]> wrote:
> Geir Magnusson wrote:
>>
>>> jvanzyl 01/10/23 22:20:55
>>>
>>> Modified: src/java/org/apache/velocity/texen/ant TexenTask.java
>>> Log:
>>> - added a fix so that contextProperties specified in the <texen>
>>> task are loaded correctly: relative paths now take ${basedir}
>>> into consideration while absolute paths are left alone. i lifted
>>> the logic recently added for loading multiple template paths :-)
>>
>> It's 8:31AM, and I just received this. Apache maillists seem a touch slow.
>>
>> I assume that this addresses Jon's scarab problems... ?
>
> I also would be interested in knowing (1) whether this is the same problem,
> or addresses the following problem:
>
> http://marc.theaimsgroup.com/?l=alexandria-dev&m=100130055026363&w=2
I honestly am not sure.
> and (2) whether or not breakages such as this are of interest to anybody in
> either the velocity or turbine or scarab communities? Previous attempts to
> find such a person were unsuccessful.
When Jon is having problems with Scarab which uses Torque and he lets me
know I usually take a look at it and we can generally get it figured it out
in a short period of time.
When Gump has a problem I admit that I usually don't go much further than
doing a clean check out and build because I don't consider Gump to be an
entirely accurate reflection of how we actually build anything. You often
have problems that we can not duplicate: what seems like disinterest to you
on my part in regard to this problem matches the disinterest I see in you in
actually having Gump run Ant the way we suggest in our docs which is to not
use shell scripts. As we have seen with the Texen tests this obviously
changes the behaviour of the build.
Please don't equate what might appear to be my disinterest in Gump output
with my reactions to a user that is having a problem. They may be the same
at some point in the future but I admit that it isn't always the case now.
Why was the issue of how we build never taken into consideration? We had
repeated problems with the texen tests which is a direct result of how Ant
changed and how Gump uses Ant. Is it not? I thought that was the case. Ant
obviously behaves differently when the classpath is constructed with a
<path> inside the build file than it does when it is constructed in the
environment in which is running Ant. There was nothing wrong with the tests
when Geir and I ran the tests on our machine but it was failing in Gump.
There was nothing wrong with Texen , the problem was build semantics. You
take many of the quirks of all the projects into consideration, why not use
the method of building that a project has chosen to adopt or maybe use both
methods to find descrepancies. Sometimes we don't find the things you find
with Gump because you build differently so what is actually accurate? The
way the project has decided to build or the way Gump builds.
That being said I am still thankful for Gump, I just don't jump sometimes
when Gump says jump. I use my judgement and whatever spare time I have left
at the end of the day.
> - Sam Ruby
--
jvz.
Jason van Zyl
http://tambora.zenplex.org
http://jakarta.apache.org/turbine
http://jakarta.apache.org/velocity
http://jakarta.apache.org/alexandria
http://jakarta.apache.org/commons
