ant-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Leonardo Abreu de Barros" <leobar...@email.com>
Subject Re: Trying to get rid of makefiles
Date Thu, 24 Jul 2003 17:04:39 GMT
First of all, I'd like to thank everyone of you for all your 
comments. Yesterday I felt like I was alone, not knowing how to go 
on, now I feel like I have hundreds of possible solutions. I still 
will take some time to study all your proposals, but at least by now 
I will answer each one of your replies, in the order they were posted:

Matthew Hawthorne: I agree with you, but I have to get this 
documentation in order to support this point of view.

David Knox: Very interesting, this information. I'll collect some 
statistics about it and gather them to a presentation I'm preparing 
showing the benefits of Ant.

David Harkness: 
- Yes, class files are mixed to java files. But I'll propose your 
solution of keeping them in separate directories.
- I agree, but the practice here seems to be always rebuild 
everything. The argument is that even when you're developing, 
whenever you change a class you risk to break some code elsewhere.
- Thanks for the suggestion, I'll take a deep look at this task.

Kevin Duffey: 
- Don't be so angry at people who resist changes. It's 
comprehensible, as the company has years of successful projects using 
makefiles, now a "newbie" (I'm here it's been 3 weeks) come with 
something nobody asked for, nobody saw the need for, and propose them 
to learn a new technology, when they already know deeply something 
else that suites to their purposes. Yes, they will finish by 
accepting changes, but it needs to be plenty justified.
- Thanks for your recommendation, I was looking for a nice book to 
acquire on this subject.
- If you consider they want to always rebuild everything, you shall 
understand why they care to every millissecond when compiling. But I 
will gather some statistics to show that we have a gain of 
performance.
- Yes, now we're using JUnit... when I arrived, a guy was studying 
it, as I had a little experience with it, I assumed this task.

Robert Anderson: yeah... I will prove it with numbers.

EJ Ciramella: 
- I totally agree with your approach, and that's the one I will take.
- I am curious about your second message: did you mean that sometimes 
you need to interrupt the build half-way, then restart later?

Conor MacNeill: rewriting may be a good idea... I just need to take 
some time to write it... but it may be the better solution.

Rutger Hofman: you've got the point... the safer approach is to send 
all the compile list to the java compiler... in other words, write my 
own version of the compiler adapter, as suggested by Conor.

Lyndon Samson: the first solution may work, but it would seem like a 
handicap of Ant being "patched" by force... the second one, using 
jikes, sounds ok to me. Even because a friend of mine told me that 
it's quite faster from the Sun's compiler.

Conor MacNeill: wow... this is REALLY hacky :)

Donald Strong: very interesting... I didn't know how this <depend> 
task worked. The <java> approach is possible too, I might figure it 
out.

Stefan Bodewig: this is really cool!! Even because of the name... I 
guess "javamake" will cause much less resistance to people used 
to "make" :)) I'll take a look at it.

Mark Evenson: As I told to Stefan, seems very nice. I will 
surely "roll up my sleeves" and take a good look at it. ;)

Thanks again,
Leonardo Barros


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@ant.apache.org
For additional commands, e-mail: user-help@ant.apache.org


Mime
View raw message