You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ant.apache.org by Peter Donald <do...@mad.scientist.com> on 2000/07/22 16:29:20 UTC
[RFE] Expansion of paths in Path task
Hi,
Is there any chance of getting some expansion occuring in the
Path/PathSet/whatever rather than relying on task doing the work. The main
reason I can see is that I want to do something like
<java ... classpath="lib/*.jar" ... />
Comments ?
Cheers,
Pete
*------------------------------------------------------*
| "Nearly all men can stand adversity, but if you want |
| to test a man's character, give him power." |
| -Abraham Lincoln |
*------------------------------------------------------*
Re: [RFE] Expansion of paths in Path task
Posted by Stefan Bodewig <bo...@bost.de>.
>>>>> "PD" == Peter Donald <do...@mad.scientist.com> writes:
PD> Hi, Is there any chance of getting some expansion occuring in the
PD> Path/PathSet/whatever rather than relying on task doing the
PD> work. The main reason I can see is that I want to do something
PD> like
PD> <java ... classpath="lib/*.jar" ... />
PD> Comments ?
There is some code in Javac that does something similar to emulate the
extension dir mechanism of JDK 1.2+ when using Jikes. The code doesn't
exactly do what Javac does as Javac doesn't care for the extension -
so we don't even need JarFileNameFilter.
Something like this could probably added to Path, you're right. I'm
going to add it to my list and revisit it if nobody gets there faster
than me.
Stefan
Re: [RFE] Incremental mode for Ant
Posted by Tom Cook <tc...@ardec.com.au>.
donaldp@mad.scientist.com writes:
> On Tue, 25 Jul 2000, James Duncan Davidson wrote:
> I still have to deal with *developers* working on 350's
> (Crazy I know but ...)
You think 350s are bad? Well, I'd better not whinge about my 300 too
much (my boss is on the list - how ya doin', Dave?) ;-)
Tom
Re: [RFE] Incremental mode for Ant
Posted by do...@mad.scientist.com.
On Tue, 25 Jul 2000, James Duncan Davidson wrote:
> Hmmmm... Or maybe just another entry point class. Instead of
> org.apache.tools.ant.Main, create a org.apache.tools.ant.LoopMain which
> called Main each time it was told to.
The problem is not so simple as you have to play with
classloader but it is certainly possible.
> Start up times are going down I think. I used to feel that pain, but either
> I've been getting used to it (I don't think so) or I'm getting faster
> machines (well, yes, I am.. :)
I still have to deal with *developers* working on 350's
(Crazy I know but ...)
Cheers,
Pete
*--------------------------------------------------*
| Latrobe University, | |
| Bundoora, Australia | Does the name 'Pavlov' |
| Office: PW220 | ring a bell ? |
| Ex: 2503 | |
*--------------------------------------------------*
Re: [RFE] Incremental mode for Ant
Posted by James Duncan Davidson <du...@x180.com>.
on 7/25/00 4:52 AM, Peter Donald at donaldp@mad.scientist.com wrote:
> At 03:09 25/7/00 -0700, you wrote:
>> How does incremental help out? -- in the M-x compile case that is..
>
> ouchies that must be painful compared to jde's way ..
To be honest, I don't do M-x compile anyways... I just run ant in an
external window and ctrl-G to goto line (or funny-key G depending on
platform and editor). It's *so* old-fashioned, I know.. :)
> Incremental mode is good because it doesn't have the loading of JVM
> everytime.
Right, I get that -- just trying to get the use case clear... If JDE keeps
something warmed up and ready to go, then by all means, a flag to support
looping compiles is a good thing...
Hmmmm... Or maybe just another entry point class. Instead of
org.apache.tools.ant.Main, create a org.apache.tools.ant.LoopMain which
called Main each time it was told to.
> I have no
> problem with the memorty overhead of a constantly in memory JVM but the
> startup time can be painful.
Start up times are going down I think. I used to feel that pain, but either
I've been getting used to it (I don't think so) or I'm getting faster
machines (well, yes, I am.. :)
But, by all means I'm not trying to block an obviously useful thing -- just
understand the particulars of the use case.
.duncan
Re: [RFE] Incremental mode for Ant
Posted by Peter Donald <do...@mad.scientist.com>.
At 03:09 25/7/00 -0700, you wrote:
>How does incremental help out? -- in the M-x compile case that is..
ouchies that must be painful compared to jde's way ..
>If you are running it from a shell window, then I can see how it would help.
Incremental mode is good because it doesn't have the loading of JVM
everytime. It becomes a larger problem when you load 6 jars as well.
Incremental mode in any way is great for rapidly fixing errors especially
when you are in a editing environment which lacks all the proper code
completion/intellisens/syntax checking that some other tools do (And yes I
know emacs has some of those features and they partially work - no where
close to comparable to some of the integrated IDE's thou). I have no
problem with the memorty overhead of a constantly in memory JVM but the
startup time can be painful.
Cheers,
Pete
*------------------------------------------------------*
| "Nearly all men can stand adversity, but if you want |
| to test a man's character, give him power." |
| -Abraham Lincoln |
*------------------------------------------------------*
Re: [RFE] Incremental mode for Ant
Posted by James Duncan Davidson <du...@x180.com>.
on 7/24/00 5:43 PM, Peter Donald at donaldp@mad.scientist.com wrote:
> At 01:42 24/7/00 -0700, you wrote:
>> This was always my thought as well. If I'm doing iterative development, I'm
>> sitting at a GUI. If I'm doing nightly builds, then a command line is fine,
>> but iterative is not a requirement.
>
> Well there is a bunch of people who use text editors like Emacs :P. It is
> very uncouth to be forced to compile via a GUI app - which removes all
> funky features Emacs does to compile lines. The only reason I want
> incremental mode is to reduce the overhead of JVM startup.
I use emacs all the time.. Well, that and a few other editors that I'm
trying out (but emacs still seems to be what I use mostly).
How does incremental help out? -- in the M-x compile case that is.. If you
are running it from a shell window, then I can see how it would help.
.duncan
Re: [RFE] Incremental mode for Ant
Posted by Peter Donald <do...@mad.scientist.com>.
At 01:42 24/7/00 -0700, you wrote:
>This was always my thought as well. If I'm doing iterative development, I'm
>sitting at a GUI. If I'm doing nightly builds, then a command line is fine,
>but iterative is not a requirement.
Well there is a bunch of people who use text editors like Emacs :P. It is
very uncouth to be forced to compile via a GUI app - which removes all
funky features Emacs does to compile lines. The only reason I want
incremental mode is to reduce the overhead of JVM startup.
Cheers,
Pete
*------------------------------------------------------*
| "Nearly all men can stand adversity, but if you want |
| to test a man's character, give him power." |
| -Abraham Lincoln |
*------------------------------------------------------*
Re: [RFE] Incremental mode for Ant
Posted by James Duncan Davidson <du...@x180.com>.
on 7/23/00 7:48 AM, Conor MacNeill at conor@m64.com wrote:
> My own thoughts were to do this in a GUI rather than in a command line. The
> basic idea was to have a GUI which attaches itself to an ant project as a
> listener and which gives colour coded output, similar to the style sheet
> supplied by Matt Foemmel. "If you want to run again, just hit the run button
> again" type of thing.
This was always my thought as well. If I'm doing iterative development, I'm
sitting at a GUI. If I'm doing nightly builds, then a command line is fine,
but iterative is not a requirement.
.duncan
Re: [RFE] Incremental mode for Ant
Posted by James Duncan Davidson <du...@x180.com>.
on 7/23/00 12:17 PM, Frederic Lavigne at fred@L2FProd.com wrote:
> I'm working on AntHill. AntHill will be available under GPL.
Um... So you wouldn't be contributing it back to this code base under the
ASF license? That's too bad. :(
.duncan
RE: [RFE] Incremental mode for Ant
Posted by Frederic Lavigne <fr...@L2FProd.com>.
Hi,
> -----Message d'origine-----
> De : Conor MacNeill [mailto:conor@m64.com]
> Envoyé : dimanche 23 juillet 2000 16:49
> À : ant-dev@jakarta.apache.org
> Objet : RE: [RFE] Incremental mode for Ant
>
>
<snip>
> I see that someone has proposed a GUI which looks kinda nice. It currently
> seems geared to building ant files rather than running them. I am also not
> sure if this code will be contributed as it seems to depend on a library,
> etc, etc.
I'm working on AntHill. AntHill will be available under GPL.
AntHill is currently able to build a selected target or the default target.
Messages (output) is directed to a specific JTextArea.
I'm encountering problem to stop the build process if ant started some exec
tasks (runtime.exec()).
I made the choice to run Ant from AntHill (in the same VM and an other
Thread).
If I start a long running target like Javadoc and I want to interrupt it, I
use theThread.interrupt();
It succesfully interrupts the thread but not the command (Process) started
by the Javadoc task.
I didn't find any way to get all the processes started by ant during the
build.
So I'm not able to interrupt it :(
> Conor
>
>
> > -----Original Message-----
> > From: Peter Donald [mailto:donaldp@mad.scientist.com]
> > Sent: Sunday, 23 July 2000 0:51
> > To: ant-dev@jakarta.apache.org
> > Subject: [RFE] Incremental mode for Ant
> >
> >
> > Hi,
> >
> > Does anyone else wnat an incremental mode for ant ? Basically what that
> > means is that you go through one run. When you get to end of run It will
> > prompt you - Enter to recompile last target, Esc to exit. This
> avoids some
> > of overhead of starting a VM and is quite a nice feature generally ... I
> > would also like a system where it was possible that I could
> type in a new
> > target to stding to get it executed when in this idle mode ..
> >
> > Cheers,
> >
> > Pete
> >
> > *------------------------------------------------------*
> > | "Nearly all men can stand adversity, but if you want |
> > | to test a man's character, give him power." |
> > | -Abraham Lincoln |
> > *------------------------------------------------------*
> >
>
RE: [RFE] Incremental mode for Ant
Posted by Conor MacNeill <co...@m64.com>.
Pete,
There was actually a request to do this a long time ago from Jonathan Locke
[jonl@eidolon.muppetlabs.com]. He included a patch which you can probably
find in the archives, although applying it now will not be an automatic
process.
My own thoughts were to do this in a GUI rather than in a command line. The
basic idea was to have a GUI which attaches itself to an ant project as a
listener and which gives colour coded output, similar to the style sheet
supplied by Matt Foemmel. "If you want to run again, just hit the run button
again" type of thing.
I see that someone has proposed a GUI which looks kinda nice. It currently
seems geared to building ant files rather than running them. I am also not
sure if this code will be contributed as it seems to depend on a library,
etc, etc.
Conor
> -----Original Message-----
> From: Peter Donald [mailto:donaldp@mad.scientist.com]
> Sent: Sunday, 23 July 2000 0:51
> To: ant-dev@jakarta.apache.org
> Subject: [RFE] Incremental mode for Ant
>
>
> Hi,
>
> Does anyone else wnat an incremental mode for ant ? Basically what that
> means is that you go through one run. When you get to end of run It will
> prompt you - Enter to recompile last target, Esc to exit. This avoids some
> of overhead of starting a VM and is quite a nice feature generally ... I
> would also like a system where it was possible that I could type in a new
> target to stding to get it executed when in this idle mode ..
>
> Cheers,
>
> Pete
>
> *------------------------------------------------------*
> | "Nearly all men can stand adversity, but if you want |
> | to test a man's character, give him power." |
> | -Abraham Lincoln |
> *------------------------------------------------------*
>
[RFE] Incremental mode for Ant
Posted by Peter Donald <do...@mad.scientist.com>.
Hi,
Does anyone else wnat an incremental mode for ant ? Basically what that
means is that you go through one run. When you get to end of run It will
prompt you - Enter to recompile last target, Esc to exit. This avoids some
of overhead of starting a VM and is quite a nice feature generally ... I
would also like a system where it was possible that I could type in a new
target to stding to get it executed when in this idle mode ..
Cheers,
Pete
*------------------------------------------------------*
| "Nearly all men can stand adversity, but if you want |
| to test a man's character, give him power." |
| -Abraham Lincoln |
*------------------------------------------------------*