makeinstall macro still allowed?

[Parag Nemade]

Hi,
  Do we allow makeinstall macro to be used in spec? I see one case
pango.spec uses it. I also see if I use the guidelines
https://fedoraproject.org/wiki/Packaging/Guidelines#Why_the_.25makeinstal...,
then we can avoid its usage.
  If FPC thinks pango package should be fixed then can someone from
FPC please commit that change?

Regards,
Parag.

[Michael Schwendt]

On Wed, 11 Aug 2010 15:21:13 +0530, Parag wrote:

>  Hi,
>    Do we allow makeinstall macro to be used in spec? I see one case
>  pango.spec uses it. I also see if I use the guidelines
>
> https://fedoraproject.org/wiki/Packaging/Guidelines#Why_the_.25makeinstal...,
>  then we can avoid its usage. 
The guidelines are clear about this. If an invocation like

  make DESTDIR=${RPM_BUILD_ROOT} install

works (and DESTDIR may have a different name), it MUST be preferred over
the brute-force %makeinstall.

On the other hand, %makeinstall (see "rpm --eval %makeinstall" for what it
does), may still be used for package, which use custom Makefiles and
don't contain a $DESTDIR installation (or a similar technique).

Also note that some packages use a shorter, expanded form of %makeinstall,
as they run something like

  make prefix=%{buildroot}%{prefix} ...

and basically do the same as %makeinstall, albeit only for some of the
variables.

>    If FPC thinks pango package should be fixed then can someone from
>  FPC please commit that change? 
It should be easy to simply give it a try first.

pango.spec doesn't follow the guidelines in other areas either, though,
e.g. the explicit Requires are missing comments.

[Ville Skyttä]

On Wednesday 11 August 2010, Michael Schwendt wrote:
>  On Wed, 11 Aug 2010 15:21:13 +0530, Parag wrote:
>  > Hi,
>  > 
>  >   Do we allow makeinstall macro to be used in spec? I see one case
>  > 
>  > pango.spec uses it. I also see if I use the guidelines
>  > https://fedoraproject.org/wiki/Packaging/Guidelines#Why_the_.25makeinstal
>  > l_macro_should_not_be_used, then we can avoid its usage.
>  
>  The guidelines are clear about this. If an invocation like
>  
>    make DESTDIR=${RPM_BUILD_ROOT} install
>  
>  works (and DESTDIR may have a different name), it MUST be preferred over
>  the brute-force %makeinstall.
>  
>  On the other hand, %makeinstall (see "rpm --eval %makeinstall" for what it
>  does), may still be used for package, which use custom Makefiles and
>  don't contain a $DESTDIR installation (or a similar technique). 
BTW, just for completeness, %make_install (with the underscore) exists 
nowadays too, and seems to do "make install DESTDIR=$RPM_BUILD_ROOT" but I 
don't see why that would be beneficial over just using the 
unobfuscated/expanded form.