[RFC/commmit,testsuite/Ada] stop using project files when building test programs

  Hello,

Below is a fairly straightforward change, except maybe for the use
of try / finally, which requires TCL version 8.6, released Dec 2012.
I think it's a reasonable requirement; any objection?

Thanks!

-------------------------------------------------------------

The current approach when building Ada programs for testing is
based on the use of a project file (testsuite/gdb.ada/gnat_ada.gpr).
To do that, we pass a number of additional arguments to target_compile,
one of them being the project file (via "-P/path/to/gnat_ada.gpr").
This used to work well-enough, but AdaCore is currently working towards
removing project-file support from gnatmake (the prefered tool for
using project files is gprbuild). So, we need to either switch
the compilation to gprbuild, or stop using project files.

First, using gprbuild is not always what users will be using to
build their applications. So having the option of using gnatmake
provides more flexibility towards exactly reproducing past bugs.
If we ever need a testcase that requires the use of gprbuild, then
I believe support for a new target needs to be added to dejagnu's
target_compile.

Also, the only real reason behind using a project file in the first
place is that we wanted to make it easy to specify the directory
where all compilation artifacts get stored. This is a consequence
of the organization choice we made for gdb.ada to keep each testcase
well organized. It is very easy to achieve that goal without using
project files.

This is therefore what this patch does: It change gdb_compile_ada
to build any program using gnatmake without using a project file
(by temporarily changing the current working directory).

There is a small (beneficial) side-effect; in the situation where
GDB is built in-tree, gnatmake is called as...

        % gnatmake [...] unit.adb

... which means that the debugging info in unit.o will say contain
a filename whose name is 'unit.adb', rather than '/path/to/unit.adb'.
This also better matches what users might typically do. But the side-
effect is that the unit name in the GDB output is not always a full
path. This patch tweaks a couple of testcases to make the path part
optional.

Note that this patch requires TCL version 8.6, which was released
Dec 2012.

gdb/testsuite:

        * lib/ada.exp (gdb_compile_ada): Reimplement avoiding the use
        of project files.
        * gdb.ada/gnat_ada.gpr: Delete.
        * gdb.ada/cond_lang.exp: Adjust test to make path before
        filename optional.
        * gdb.ada/small_reg_param.exp: Likewise.

Tested on x86_64-linux, with both in-tree and out-of-tree builds.
---
 gdb/testsuite/gdb.ada/cond_lang.exp       |  2 +-
 gdb/testsuite/gdb.ada/gnat_ada.gpr        | 25 -------------------------
 gdb/testsuite/gdb.ada/small_reg_param.exp |  2 +-
 gdb/testsuite/lib/ada.exp                 | 20 ++++++++++++++++----
 4 files changed, 18 insertions(+), 31 deletions(-)
 delete mode 100644 gdb/testsuite/gdb.ada/gnat_ada.gpr
  

On 12/22/2015 03:33 PM, Joel Brobecker wrote:
> Hello,
> 
> Below is a fairly straightforward change, except maybe for the use
> of try / finally, which requires TCL version 8.6, released Dec 2012.
> I think it's a reasonable requirement; any objection?

I'm still on F20, and that shipped with 8.5 too.  I'll have to upgrade
eventually, so this may the trigger.

Note that requiring newer expect/tcl may mean that fewer people will do
the occasional gcc compile farm testing on some older hosts.  From a quick
look at some machines there, I see:

 gcc20 (debian wheezy)'s expect links with 8.5.
 gcc111 (AIX 7.1)'s expect links with 8.4.
 gcc110 (F18 POWER7/ppc64)'s expect links with 8.5.

How hard would it be to avoid try/finally?  Wouldn't you just have to
use catch instead?

Are there other nice 8.6 features that it'd be really nice to
make use of too?

Thanks,
Pedro Alves
  

> > Below is a fairly straightforward change, except maybe for the use
> > of try / finally, which requires TCL version 8.6, released Dec 2012.
> > I think it's a reasonable requirement; any objection?
> 
> I'm still on F20, and that shipped with 8.5 too.  I'll have to upgrade
> eventually, so this may the trigger.
> 
> Note that requiring newer expect/tcl may mean that fewer people will do
> the occasional gcc compile farm testing on some older hosts.  From a quick
> look at some machines there, I see:
> 
>  gcc20 (debian wheezy)'s expect links with 8.5.
>  gcc111 (AIX 7.1)'s expect links with 8.4.
>  gcc110 (F18 POWER7/ppc64)'s expect links with 8.5.

Humpf, that's a fair number of reasons showing that assuming 8.6
may not be reasonable. Bouh...

> How hard would it be to avoid try/finally?  Wouldn't you just have to
> use catch instead?

I don't think it would be very hard. I think catch will work, but
will be a little more convoluted. I'll give it a try...

> Are there other nice 8.6 features that it'd be really nice to
> make use of too?

This is what the 8.6 release notes say:
(https://www.tcl.tk/software/tcltk/8.6.html)

Highlights of Tcl 8.6

  * Object Oriented Programming
       [prob not very intesting for us, but see URL if interested]

  * Stackless Evaluation:
       The evaluation of many levels of nested proc calls are no longer
       implemented as a stack of nested C routine calls. This revision
       in the internal implementation of Tcl evaluation makes deep
       recursion in Tcl scripts safe to do. But there's more...

       This new implementation enables a collection of new commands,
       coroutine, tailcall, yield, and yieldto that provide profound new
       capabilities and models of concurrency to Tcl scripts.

  * Enhanced Exceptions:
       New commands try and throw and a wealth of new -errorcode values
       enable far more precise trapping and handling of exceptions using
       a familiar construct.

  * Batteries Included:
       Tcl delivers in the pkgs subdirectory a bundled collection of
       third-party packages built and installed along with Tcl.

  * Thread-enabled Operations:
       A thread-enabled default build, a bundled Thread package, and new
       command interp cancel make Tcl 8.6 ready for your multi-threaded
       programming tasks.

  * SQL Database Powered:
       [prob not very intesting for us, but see URL if interested]

  * IPv6 Networking:
       Both client and server sockets support IPv6 where platform
       support exists.

  * Built-in Zlib Compression:
       New command zlib provides utilities to handle compression of data
       and streams.

  * List Processing:
       New commands lmap and dict map enable the elegant expression of
       transformations over Tcl containers.

  * Stacked Channels by Script:
       New commands chan push and chan pop expose the power of stacked
       channels without the need to write C code.

  * Additional New Features:
       Temporary file creation, enhancements to list sorting and
       setting, dict filtering, half-close of bidirectional channels,
       encoding and decoding of binary sequences, finer control over
       load, and many many more.

Maybe some interesting stuff, but I wouldn't want compilation of
Ada code be the trigger that breaks people's testing. Upgrading
tcl/expect is not always a trivial task...

Thanks for the feedback,