[PATCH] Starteam Checkout

[Samuel R Listopad II <sl...@intercapsystems.com>]

This patch attempts the following:

	Allows the starteam checkout to use Status information to filter what
should be checked out.
		A String of the form "CUGI", definition of chars in javaDoc.
	Adds an appendViewName attribute to allow one to not have the View Name
appended (default is append,
		which is what current incarnation does)
	Updates the javaDoc to be more accurate, and explain new features.

Sam

RE: [PATCH] Starteam Checkout

[Jon Tirsen <jo...@emm.com>]

There are two problems to the StarTeam checkout-task (that's why I built my
own instead, of using this one):
1) The way of specifying all the different properties of the connection is
not very idiomatic to StarTeam, one usually specifies the connection using a
StarTeam-URL that has a general form. (Don't have it in my head but I can
look it up if you like.)
2) The StarTeam-API has a bug that shows up when you check out a large
number of files. It has to do with executing the "touch" command. It's much
better to use the checkOutAsInputStream (or whatever the name was of that
method). Basically the check-out ends with a core-dump from the VM, which is
a bit annoying... :-) In Java 1.2 there is an API for changing last-modified
date. (But it won't work in 1.1 of course.) You might want to provide an
option for that (or make it the default, or something).

-----Original Message-----
From: Stefan Bodewig [mailto:bodewig@bost.de]
Sent: Monday, November 06, 2000 6:19 AM
To: ant-dev@jakarta.apache.org
Subject: Re: [PATCH] Starteam Checkout


Samuel,

could you please rewrite your patch in a way that it compiles on JDK
1.1 (no HashMap please). And then I'd suggest to make
setAppendViewName take a boolean argument.

Finally I'm a little hesitant to commit a patch I cannot test myself,
any Starteam users apart from Samuel here that could check that it
works in their environment as well?

Stefan

Re: Revised: [PATCH] Starteam Checkout

[Stefan Bodewig <bo...@bost.de>]

Samuel R. Listopad, II <sl...@intercapsystems.com> wrote:

>> But it works for "not too many" files, so it's not worse than what
>> we already have?
> 
> Some changes: Only change modified date if changeLastModified
> variable set, and on 1.2 (won't work on 1.1, though it is written
> with reflection so it will compile on 1.1, is this acceptable?)

Sure, this is what touch, untar and other tasks do as well.

> But as I said Jon is going to test it again before I resubmit, so if
> you could hold off on the commit til then

OK, no problem, just wanted to shorten my patch backlog.

Stefan

RE: Revised: [PATCH] Starteam Checkout

[Samuel R Listopad II <sl...@intercapsystems.com>]

> But it works for "not too many" files, so it's not worse than what we
> already have?

Yes, however I have revamped it again and am talking to Jon right now to see
if it fixes the problems.
Some changes:
	Only change modified date if changeLastModified variable set, and on 1.2
(won't work on 1.1, though it is written with reflection so it will compile
on 1.1,  is this acceptable?)

	Check out via stream and update status by resync, rather than just using
thier implementation of checkout (which caused the seg fault on solaris)

But as I said Jon is going to test it again before I resubmit,  so if you
could hold off on the commit til then

Sam

Re: Revised: [PATCH] Starteam Checkout

[Stefan Bodewig <bo...@bost.de>]

Jon Tirsin <jo...@emm.com> wrote:

> I've tested it and it's not working (a core-dump if you check out to
> many files on Solaris).

But it works for "not too many" files, so it's not worse than what we
already have?

Stefan

RE: Revised: [PATCH] Starteam Checkout

[Jon Tirsén <jo...@emm.com>]

I've tested it and it's not working (a core-dump if you check out to many
files on Solaris). It's a previously known bug with the StarTeam SDK.
There is a work-around but that requires 1.2.

Also, an RFE is to add support for StarTeam URL's instead.


-----Original Message-----
From: Samuel R Listopad II [mailto:slistopad@intercapsystems.com]
Sent: Monday, November 06, 2000 12:19 PM
To: ant-dev@jakarta.apache.org
Subject: Revised: [PATCH] Starteam Checkout



> could you please rewrite your patch in a way that it compiles on JDK
> 1.1 (no HashMap please).

Done.

> And then I'd suggest to make
> setAppendViewName take a boolean argument.

And Done.

> any Starteam users apart from Samuel here that could check that it
> works in their environment as well?

Yeah if someone could do some testing of this patch I would be greatly
appreciative.
Any questions, bugs or concerns should be sent to the list and me if
possible.

Sam

Revised: [PATCH] Starteam Checkout

[Samuel R Listopad II <sl...@intercapsystems.com>]

> could you please rewrite your patch in a way that it compiles on JDK
> 1.1 (no HashMap please).

Done.

> And then I'd suggest to make
> setAppendViewName take a boolean argument.

And Done.

> any Starteam users apart from Samuel here that could check that it
> works in their environment as well?

Yeah if someone could do some testing of this patch I would be greatly
appreciative.
Any questions, bugs or concerns should be sent to the list and me if
possible.

Sam

Re: [PATCH] Starteam Checkout

[Stefan Bodewig <bo...@bost.de>]

Samuel,

could you please rewrite your patch in a way that it compiles on JDK
1.1 (no HashMap please). And then I'd suggest to make
setAppendViewName take a boolean argument.

Finally I'm a little hesitant to commit a patch I cannot test myself,
any Starteam users apart from Samuel here that could check that it
works in their environment as well?

Stefan