ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From James Duncan Davidson <dun...@x180.net>
Subject Re: cvs commit: jakarta-ant/proposal/anteater/lib crimson.jar jaxp.jar
Date Wed, 06 Dec 2000 17:28:19 GMT
On 12/6/00 2:14 AM, "Peter Donald" <donaldp@apache.org> wrote:

> At 09:58  6/12/00 -0000, you wrote:
>>   public class Bootstrap {
>>      
>>       private static String base = "../";
>>  +    private static String crimsonSources = "../../../xml-crimson/src";
> // relative to base
> 
> This makes it near impossible for a lot of people to build it with this
> tool. 

Why? If you are checking out jakarta-ant, it's just another word on the
checkout command to checkout crimson.jar.

> It certainly up the stakes even for those who are willing to spend
> the extra time/effort to check it out - worse it could leads us to not
> being able to determine whos at fault after a change (ie crimson people
> make a mistake and it could effect you). I dislike spiderweb builds for
> this very reason.

Crimson (or any other XML parser) changes can affect us at any time even if
we are using "promoted" jars. In fact, if we use promoted jars, then we
don't know that Crimson busted something stupid till whenever we decide to
update (whenever that might be). However, if you make things dependant, you
know for sure whenever the depended on tool busts -- which makes identifying
the culprit a bit easier.

So, I'd prefer a CJAN, or a setup with a central set of Jars to doing
this... But for right this second, given the balance of options, this avoids
having yet another jar in the repository. And for sure checking in a
crimson.jar into the workspace means that that jar will sit forever and then
crimson will change and then something will really be screwed up -- which is
the situtation Stylebook and so many other tools are in right now.

-- 
James Duncan Davidson                                        duncan@x180.net
                                                                  !try; do()


Mime
View raw message