Re: gtk3 branch fails to compile

[Andreas Schmidt <pi-c arcor de>]

Hi, Peter,

thanks for your prompt help!

On 01/13/2017 03:23:05 PM, Peter Bloomfield wrote:
> Hi Andreas,
>
> On 01/13/2017 07:27:20 AM Fri, Andreas Schmidt wrote:
> > Hi,
> >
> > I have a similar problem.
> >
> > On 11/29/2016 08:47:22 AM, Helmut Jarausch wrote:
> > > Hi,
> > >
> > > probably after the update to gtk+-3.22.4 the  GIT-version of Balsa  
> > > (gtk3 branch) fails to compile:
> > The last version of balsa I could compile (using this now) is  
> > 2.5.2-71-gd032986. Meanwhile, GTK+ is at version 3.22.6-1  
> > (Debian/sid), and I just tried compiling balsa from the current  
> > master branch.
> >
> > > libbalsa.c: In function 'ask_cert_real':
> > > libbalsa.c:564:5: error: 'gtk_window_set_wmclass' is deprecated  
> > > [-Werror=deprecated-declarations]
> > >
> > > libbalsa.c: In function 'ask_timeout_real':
> > > libbalsa.c:611:5: error: 'gtk_window_set_wmclass' is deprecated  
> > > [-Werror=deprecated-declarations]
>
> Those errors were fixed (I hope!) with commit  
> 1a894f5f0e2776d6ad963bf5a6b4d2aa708ee8e8.
I guess so -- at least I didn't get these errors.

> > Different output here:
> >
> > ***
> > imap-tls.c:113:1: error: ‘dyn_destroy_function’ defined but not used  
> > [-Werror=unused-function]
> >  dyn_destroy_function(struct CRYPTO_dynlock_value *l,
> >  ^~~~~~~~~~~~~~~~~~~~
> > imap-tls.c:103:1: error: ‘dyn_lock_function’ defined but not used  
> > [-Werror=unused-function]
> >  dyn_lock_function(int mode, struct CRYPTO_dynlock_value *l,
> >  ^~~~~~~~~~~~~~~~~
> > imap-tls.c:91:1: error: ‘dyn_create_function’ defined but not used  
> > [-Werror=unused-function]
> >  dyn_create_function(const char *file, int line)
> >  ^~~~~~~~~~~~~~~~~~~
> > imap-tls.c:80:1: error: ‘id_function’ defined but not used  
> > [-Werror=unused-function]
> >  id_function(void)
> >  ^~~~~~~~~~~
> > imap-tls.c:71:1: error: ‘locking_function’ defined but not used  
> > [-Werror=unused-function]
> >  locking_function(int mode, int n, const char *file, int line)
> >  ^~~~~~~~~~~~~~~~
> > cc1: all warnings being treated as errors
> > ***
>
> Pinning down exactly when OpenSSL introduced that problem is proving  
> difficult--please try the attached patch.
The patch seems to work -- no errors this time!

> > The pattern is the same, though: Warnings are treated as errors (how  
> > can I stop that?), and therefore, the build fails. I have attached  
> > the logfile from make.
>
> Building in a git tree turns warnings into errors, and Balsa detects  
> that you are building in a git tree by the presence of a .git  
> subdirectory. If you rename it, warnings will be left as warnings  
> (but git will not recognize it as a git tree!).
Great! Before applying the patch, I tried renaming the .git  
subdirectory and balsa compiled without problems. However, I just  
noticed something different: When I compile balsa without .git,  
Balsa-->About shows the expected version number (2.5.3). Now, with the  
subdir present again, I get version 2.5.2-357-g0f6c528a. I guess, the  
suffix is necessary for git, but shouldn't the version number start  
with 2.5.3, too?

Best regards

Andreas

Attachment: pgpNQESk7d_ql.pgp
Description: PGP signature