commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Libbrecht <p...@activemath.org>
Subject Re: [jelly] distribution format
Date Tue, 14 Jun 2005 08:40:57 GMT

I'll try to help directly and keep things moving for this week.
I hence propose to:

- change the script to "$@", I am positive moving to "$1" "$2" is what 
you wish (the quote gets removed by the script-invocation, see PS for a 
test)
- drop or not drop classpath whichever you wish.
- also, in the archive, as noted:
   - the scripts are in Windows end-of-lines
   - the x-bit in the tar.gz for the bin/jelly is missing (and in the 
zip if possible, ant allows that)

- make a README for binaries saying which taglibs are installed and 
that extensibility is either done by hand following dependencies or 
using src.
An example I made is at:
	http://people.apache.org/~polx/README-binary-jelly.txt

Brett, I really hope not to kill your efforts! Sorry to look so 
"constraining". Do tell me if I should either commit this or let you do 
so so that it streams in the releases...

Btw, how can, in a vote, be answered: "yes with the changes made 
already" as opposed to the release-candidates presented in the home 
page ???

paul

PS: here's a test for $* vs "$@":
<j:jelly xmlns:j="jelly:core">
   Arg[0] is ${args[0]}.
   Arg[1] is ${args[1]}.
   Arg[2] is ${args[2]}.
</j:jelly>
If invoked with binary-delivered bin/jelly with the parameters "a b" 
"c" it gives the output:
: Arg[0] is /tmp/blop.jelly.
:  Arg[1] is a.
:  Arg[2] is b.
If the bin is changed to "$@" you get the output:
Arg[0] is /tmp/blop.jelly.
   Arg[1] is a b.
   Arg[2] is c.


Le 14 juin 05, à 01:59, Brett Porter a écrit :

> Paul - thanks for your feedback, comments below. Can I ask you to 
> formally vote on the vote thread, too - do these comments lead to a -1 
> (hold the release until fixed), -0 (prefer them fixed, but not 
> essential) or +1 (these can wait until next time) ? This *has* to be 
> on the [vote] thread.
>
> I'll also point out I have limited time to work on this now, and 
> starting with JavaOne I'll be away for 3 weeks - so either this is out 
> this week, someone else steps up to finish it, or it waits until 
> August. I'm not in anyway trying to influence the vote here - just to 
> give necessary perspective and make sure we keep moving.
>
> Paul Libbrecht wrote:
>
>> My first comments on the binary:
>> - packaging is simple and straightforward, that's good!
>> - last line of the sh script should have "$@" around instead of $* I 
>> believe
>>   (otherwise you don't allow spaces in parameters, should I commit 
>> this?)
>
> I thought it was correct ("$@" expands to "$1" "$2" ... which I don't 
> think works when $1 is already quoted, but I may be wrong).
>
>> - do we not want to include an "endorsed" directory since it would 
>> allow to circumvent the shipped parser and a possibly too old version 
>> of xalan??
>
> ... or require 1.4 for the standalone and get rid of all of them and 
> halve the distribution size :)
>
> In the current situation, I think the JDK 1.4 or JDK 5.0 parser will 
> be used which is probably a good thing (at least with 5.0 it has a 
> newer Xerces built in). From what I can tell, Jelly works just fine 
> under the built in parsers of these JDKs.
>
>>
>> - I would resign to take in account an existing CLASSPATH variable in 
>> the script ... it tends to add unpredictability. People can still 
>> hack the script if they wish.
>
> I don't really mind either way.
>
>> - the README is for the source... we need to have one for the binary, 
>> or?
>>   It should include:
>>   - which taglibs are included (not sure of the list, looks like xml 
>> is not for example)
>>   - which examples can be run
>>   --> about this, I feel we should include examples, or maybe 
>> examples with URLs ??
>>   - how to download (and install) more taglibs (if possible)
>>   --> about this, I fear forehead will not support this... why not 
>> use shell
>>    script to define the classpath using a command that takes all jars 
>> in the
>>    lib directory ? (or let it be with extensibility and refer to the 
>> source for it,
>>    not very nice)
>
> This seems reasonable, though I'd hope the site gives them enough info 
> and is easy enough to find. I think the one README can be used for 
> both, starting with the binary instructions.
>
> But really, I don't have the time to spend on this, this week.
>
>>
>> ... just to catch up quickly on talk exchanges about the 
>> documentation inclusion: I think we should include the produced 
>> web-site in both the source and binary distribution since it cannot 
>> be built with either of them.
>
> You can build the documentation in the source with "maven xdoc". 
> Perhaps the README should be adjusted accordingly.
>
>>
>> Sorry for the late replies.
>
> Better late than never! :)
>
> Cheers,
> Brett
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: commons-dev-help@jakarta.apache.org
>


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-dev-help@jakarta.apache.org


Mime
View raw message