Re: [PATCH] Introduce "gdb/configure.nat" (and delete "gdb/config/*/*.mh" files)

[Sergio Durigan Junior <sergiodj@redhat.com>]

On Tuesday, May 02 2017, Simon Marchi wrote:

> On 2017-05-02 15:28, Sergio Durigan Junior wrote:
>>>> +# Variables defined here:
>>>> +#
>>>> +# NAT_FILE
>>>> +# NATDEPFILES
>>>> +# NAT_CDEPS
>>>> +# LOADLIBES
>>>> +# MH_CFLAGS
>>>> +# XM_CLIBS
>>>> +# NAT_GENERATED_FILES
>>>> +# HAVE_NATIVE_GCORE_HOST
>>>> +# NAT_EXTRA_FRAGS_FILE
>>>
>>> Could you document (if you know it) what each variable does?
>>
>> Sure.  Here's what I know what them.  What do you think?
>>
>> # Variables defined here:
>> #
>> # NAT_FILE - The header file with definitions for this host
>
> Looks good.
>
>> #
>> # NATDEPFILES - The depfiles to be compiled for this host
>
> "depfiles" is not so clear I think.  Perhaps
>
>   NATDEPFILES - Source files required for native debugging on this host.

Adopted.

>> #
>> # NAT_CDEPS - Dynamic symbols to be exported for libthread_db
>
> Looks good.
>
>> #
>> # LOADLIBES - Libraries that will be linked against GDB for this host
>
> Don't we usually say the reverse, the program is linked against the
> libraries?
>
>   LOADLIBES - Libraries against which GDB will be linked for this host.

You're right.  Fixed.

>> #
>> # MH_CFLAGS - Additional CFLAGS for this host
>
> Looks good.
>
>> #
>> # XM_CLIBS - Host-dependent libs to be linked against GDB.
>
> Same comment about "linked against".

Fixed.

>> # NAT_GENERATED_FILES - Generated files by this host
>
> by -> for?
>
>   NAT_GENERATED_FILES - Generated files for this host.

Right, fixed.

>> # HAVE_NATIVE_GCORE_HOST -
>
> I just noticed that this variable is not defined in Makefile.in.
> configure.ac AC_SUBST's _TARGET twice, but not _HOST, so I think you
> confused them.

Yeah, I did confuse.  Fixed on gdb/Makefile.in.

> But to be honest, I am also completely confused by what those two
> variables do.

So, doing a bit of archaeology I found that these flags were added by
b292c7831823605b02a564b0a9274f95141c92ee.  According to the discussions
relevant to this commit
(<https://sourceware.org/ml/gdb-patches/2013-04/msg00321.html>), I see
that these flags serve to signalize whether the gcore script will be
installed or not.

>> # NAT_EXTRA_FRAGS_FILE - Extra Makefile fragmentsFile containing
>> extra fragments of Makefile
>> #                        that will be used by this host.
>
> That could be shortened to
>
>   NAT_EXTRA_FRAGS_FILE - Extra Makefile fragments that will be used
> for this host.

Hm, not sure.  My intention was to make it clear that this variable
needs to contain a filename.  Your proposed description could be
interpreted differently, I'm afraid.

> Make sure they all end (or don't end) with a period, for consistency.

Sure; this was just a quick draft, I fixed the inconsistencies now :-).

> Perhaps someone could shed some light on the obscure variables names?
> What do the XM_ and MH_ prefixes stand for?  What's the difference
> between XM_CLIBS and LOADLIBES (and why is it called "LIBES"?) ?

Good question, I was also wondering that myself :-).

>>> IMO, the interest of having all of this in a single file is to be able
>>> to factor out common things.  A lot of NATDEPFILES are repeated Would
>>> it be possible to have a switch on ${gdb_host} at the top level, and
>>> specify all the files specific to OSes but machine-agnostic?  For
>>> example, fork-child.o and inf-ptrace.o probably appear in all the
>>> linux ports.
>>
>> Fair point.  I will address what you and Jon mentioned, and resubmit
>> the
>> patch with the modifications.
>
> Ok.  But as John mentioned, it's probably better if this patch is a
> simple translation of what's already there, and any "optimization"
> coming in a subsequent patch.  It will be easier to track the changes.

Sure.  I'll send a series of 2 patches, the first one containing the
original patch (+ the proposed modifications by you and Pedro), and the
second one containing the rearrangement to make things clearer.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/