Build failure from trunk

Build failure from trunk

[Matthew Leach]

Hi Emacs,

I'm having issues building Emacs after a recent update. I'm now getting:

/home/matthew/.emacs_imp/src/xfns.c:4322: undefined reference to `x_get_net_workarea'

When I revert the revision: eggert@cs.ucla.edu-20130507053401-0nu6s4p8gb1fvsx0

this fixes the problem.

Thanks,
Matt

Re: Build failure from trunk

[Jan Djärv]

Hello.

7 maj 2013 kl. 10:43 skrev Matthew Leach <matthew@mattleach.net>:

> Hi Emacs,
> 
> I'm having issues building Emacs after a recent update. I'm now getting:
> 
> /home/matthew/.emacs_imp/src/xfns.c:4322: undefined reference to `x_get_net_workarea'
> 
> When I revert the revision: eggert@cs.ucla.edu-20130507053401-0nu6s4p8gb1fvsx0
> 
> this fixes the problem.

One would hope people would at least compile before checking in, but apparently not.
Trunk is fixed now.

	Jan D.

Re: Build failure from trunk

[Paul Eggert]

On 05/07/13 10:48, Jan Djärv wrote:
> One would hope people would at least compile before checking in

Sorry about that; but I did compile, and it worked on my platform.
Unfortunately I didn't notice one of the ifdefs that didn't trigger
for my platform.  I checked in a further version that should work
for both kinds of platforms.

Re: Build failure from trunk

[Jan Djärv]

Hello.

> Sorry about that; but I did compile, and it worked on my platform.
> Unfortunately I didn't notice one of the ifdefs that didn't trigger
> for my platform.  I checked in a further version that should work
> for both kinds of platforms.

#if !defined USE_GTK || GTK_MAJOR_VERSION < 3 + (GTK_MINOR_VERSION < 4)

This will be true for Gtk 4.0 to 4.3, 5.0 to 5.3, 6.0 to 6.3 and so on, so it is obviously not correct.

	Jan D.

Re: Build failure from trunk

[Paul Eggert]

On 05/07/2013 10:06 PM, Jan Djärv wrote:
> #if !defined USE_GTK || GTK_MAJOR_VERSION < 3 + (GTK_MINOR_VERSION < 4)
> 
> This will be true for Gtk 4.0 to 4.3, 5.0 to 5.3, 6.0 to 6.3 and so on, so it is obviously not correct.

Sorry, I don't see the bug.  Perhaps you misread the "+" as "|"?
For 4.0, for example, GTK_MAJOR_VERSION is 4 and GTK_MINOR_VERSION is 0,
so the comparison evaluates to 4 < 3 + (0 < 4),
which is false, which is what is wanted here.

Re: Build failure from trunk

[Jan Djärv]

Hello.

8 maj 2013 kl. 16:46 skrev Paul Eggert <eggert@cs.ucla.edu>:

> On 05/07/2013 10:06 PM, Jan Djärv wrote:
>> #if !defined USE_GTK || GTK_MAJOR_VERSION < 3 + (GTK_MINOR_VERSION < 4)
>> 
>> This will be true for Gtk 4.0 to 4.3, 5.0 to 5.3, 6.0 to 6.3 and so on, so it is obviously not correct.
> 
> Sorry, I don't see the bug.  Perhaps you misread the "+" as "|"?
> For 4.0, for example, GTK_MAJOR_VERSION is 4 and GTK_MINOR_VERSION is 0,
> so the comparison evaluates to 4 < 3 + (0 < 4),
> which is false, which is what is wanted here.

I mixed up the precedence of < and +.  Anyway, this is more unclear that the corresponding #if later in the code, i.e.:

#if GTK_MAJOR_VERSION > 3 || (GTK_MAJOR_VERSION == 3 && GTK_MINOR_VERSION >= 4)

At least they should be the same so someone reading the code can see that they are related.
We don't have that style when checking Gtk+ versions in other places, so introducing it here just makes the code harder to read.

	Jan D.

Re: Build failure from trunk

[Paul Eggert]

On 05/08/13 09:01, Jan Djärv wrote:
> At least they should be the same so someone reading the code can see that they are related.

Good point.  Glibc has a good style here; it defines macros
__GNUC_PREREQ, __GLIBC_PREREQ, etc., to check version numbers.
I added a macro GTK_PREREQ to do that with the similar uses
for GTK version numbers, as trunk bzr 112523.  Please feel
free to change it if you still find it confusing.

Re: Build failure from trunk

[Jan Djärv]

Hello.

9 maj 2013 kl. 01:00 skrev Paul Eggert <eggert@cs.ucla.edu>:

> On 05/08/13 09:01, Jan Djärv wrote:
>> At least they should be the same so someone reading the code can see that they are related.
> 
> Good point.  Glibc has a good style here; it defines macros
> __GNUC_PREREQ, __GLIBC_PREREQ, etc., to check version numbers.
> I added a macro GTK_PREREQ to do that with the similar uses
> for GTK version numbers, as trunk bzr 112523.  Please feel
> free to change it if you still find it confusing.

Good idea.  But you missed one in gtkutils.c.

	Jan D.

Re: Build failure from trunk

[Paul Eggert]

On 05/09/2013 01:10 AM, Jan Djärv wrote:
> Good idea.  But you missed one in gtkutils.c.

I also missed the fact that GTK already defines a version-checking
macro.  I fixed both problems in trunk bzr 112533.

Re: Build failure from trunk

[Jan Djärv]

Hello.

9 maj 2013 kl. 16:51 skrev Paul Eggert <eggert@cs.ucla.edu>:

> On 05/09/2013 01:10 AM, Jan Djärv wrote:
>> Good idea.  But you missed one in gtkutils.c.
> 
> I also missed the fact that GTK already defines a version-checking
> macro.  I fixed both problems in trunk bzr 112533.

Nice.

	Jan D.