[tentacles] assertions

[Robert Burrell Donkin <ro...@blueyonder.co.uk>]

Tentacles is a little unusual in using Java assertions. When assertions 
are off, then [1] no longer works as expected (rather than running with 
debugging checks off).

For applications in runnable jars, I think using assertions to allow 
debugging is a cool idea - controlling logging levels is sometimes 
tricky in this situation. However - by the principle of least surprise - 
when on the golden path, I think tentacles should function correctly 
whether assertions are on or off.

Opinions?

Robert
[1] 
http://svn.apache.org/viewvc/creadur/tentacles/trunk/src/main/java/org/apache/creadur/tentacles/Files.java?revision=1355691&view=markup

Re: [tentacles] assertions

["P. Ottlinger" <po...@aiki-it.de>]

Am 29.03.2013 10:06, schrieb Robert Burrell Donkin:
> I've been wondering whether Tentacles might be a good GSOC project for a
> student who's interested more in solving human problems than hard core
> technical coding. Perhaps run as a Agile project, weekly sprints going
> back to people in the Incubator and other projects for new features.
> Opinions welcomed :-)

+1

This way we could determine new features and develop some sort of roadmap.

Cheers
Phil

Re: [tentacles] assertions

[Robert Burrell Donkin <ro...@blueyonder.co.uk>]

On 03/28/13 18:00, David Blevins wrote:

<snip>

> And I wrote the stinking code.

I feel the same about the original Rat code base :-) And those design 
experiments still hurt Rat today...

So with Tentacles, probably worthwhile tidying up and cleaning up before 
pushing onwards

<snip>

> Feel free to hack it up.

:-)

> Side note, wow I didn't know anyone but be had ever used tentacles :)

I have some scripts which do similar stuff, so having Tentacles here has 
given me the impetus to start working on better solutions :-)

I've been wondering whether Tentacles might be a good GSOC project for a 
student who's interested more in solving human problems than hard core 
technical coding. Perhaps run as a Agile project, weekly sprints going 
back to people in the Incubator and other projects for new features. 
Opinions welcomed :-)

Robert

Re: [tentacles] assertions

[David Blevins <da...@gmail.com>]

On Tue, Mar 26, 2013 at 10:58 AM, Robert Burrell Donkin
<ro...@blueyonder.co.uk> wrote:
> Tentacles is a little unusual in using Java assertions. When assertions are
> off, then [1] no longer works as expected (rather than running with
> debugging checks off).
>
> For applications in runnable jars, I think using assertions to allow
> debugging is a cool idea - controlling logging levels is sometimes tricky in
> this situation. However - by the principle of least surprise - when on the
> golden path, I think tentacles should function correctly whether assertions
> are on or off.
>
> Opinions?

Totally agree.  I've never once remembered to turn them on, have
always ran into the issue of the dir not getting created, and *then*
remembered about assertions needing to be on.

And I wrote the stinking code.

I rarely used assertions before and have completely stopped using them
since.  It's a clear no-no for code that must run, and really, who
wants even their checks to be optional?

Feel free to hack it up.

Side note, wow I didn't know anyone but be had ever used tentacles :)

-David

Re: [tentacles] assertions

[sebb <se...@gmail.com>]

On 26 March 2013 09:58, Robert Burrell Donkin
<ro...@blueyonder.co.uk> wrote:
> Tentacles is a little unusual in using Java assertions. When assertions are
> off, then [1] no longer works as expected (rather than running with
> debugging checks off).
>
> For applications in runnable jars, I think using assertions to allow
> debugging is a cool idea - controlling logging levels is sometimes tricky in
> this situation. However - by the principle of least surprise - when on the
> golden path, I think tentacles should function correctly whether assertions
> are on or off.
>
> Opinions?

That use of assert is clearly a bug because it has side effects.

The return code from mkdirs() should have been saved in a variable,
which is then tested by the assert.

Maybe it was that way once and someone inadvertently inlined the code.

If so - and to prevent it happening again - the variable should be
documented as being required.

> Robert
> [1]
> http://svn.apache.org/viewvc/creadur/tentacles/trunk/src/main/java/org/apache/creadur/tentacles/Files.java?revision=1355691&view=markup