Re: GNATBench : how can I install ?

[Britt Snodgrass <britt.snodgrass@gmail.com>]

On Nov 4, 11:08 pm, Yannick Duchêne (Hibou57)
<yannick_duch...@yahoo.fr> wrote:
> Le Thu, 04 Nov 2010 20:08:53 +0100, Britt Snodgrass
> <britt.snodgr...@gmail.com> a écrit:
>
> > Forget Hibachi, its a (effectively) dead project.
>
> Some already told me the same here, but I could not get an idea without a
> try. Hibachi is not so bad, it is lightweight and simple, which can be
> good. And I don't like to say “dead” for such a thing which was not
> created on that purpose (some care for the authors too). This was still a
> good and not-tiny job.

Yes, it has some good qualities but I don't see the point of using it
when GNAT is your compiler. GNATbench works much better with GNAT.
Hibachi probably works better with an Aonix compiler.  I actually used
earlier versions of the Aonix ADT before it was released as Hibachi.
The company I work for funded some of the initial Aonix ADT
development because we needed an Ada plug-in for Eclipse and couldn't
get AdaCore interested at the time (several years ago).  I think
that's the only reason Hibachi has some support for GNAT. We switched
to GNATbench starting with version 2.1 and now use GNATbench 2.4.


>
> I may add I've checked it works fine with MinGW, but still requires a
> minimal GPS in the path… to open the GPR file dialog box.

I didn't think there was any remaining requirement to also have GPS
installed but I haven't checked recently.  I always keep the latest
release of GPS installed as well.  I use both GPS and GNATbench and
like them both.  They each have their strengths, and its very easy to
switch between them, or even run them both at the same time since they
can "share" a GPR file.

>
> > Version 2.5.0 of
> > GNATbench (when released) will have a more standard "local update
> > site" installation mechanism.
>
> Seems they may have reach this target while they may not tell it : there
> is even no need to use the installer, I've played a bit to see how it
> works. All you have to do (at least in Windows), is to unpack the
> installer or the archive, move the “plugins” and “features” folders in
> your eclipse directory (nothing else at all), then launch Eclipse.

Yes there are various ways to install it. I just described the
standard way in my other post.

>
> All that means it is near to be possible to install it straight-away (it
> just needs to get ride of the minimal external GPS it relies on, to be OK).
>
> What is this “gnatbench.link” ? Mine is working without. I tried the
> installation from place to place to see if this was still standalone… and
> it is.

Eclipse can load plug-ins from various locations if it finds a link
file containing the plug-in path. With the default installation
method, GNATbench 2.4 (and earlier) will create the gnatbench.link in
the Eclipse links folder if you install GNATbench in the GNAT
directory.

>
> Compared to Hibachi, I could not strictly say one is better than the
> other, this would be like comparing AdaGIDE and GPS (no way to compare,
> the spirit and target are not the same). Both  have their weakness (CDT on
> the other hand is very good… now he know a possible reason why so much
> people are sticked to C/C++/Java… the environment is very neat-good in
> this area, and all seamless).
>

GNATbench is good and getting better with each release.  It also has
built-in support for running the SPARK tools (as does GPS).

Hibachi is an open project and could be improved, but I don't see that
anyone is working on it.