ROOT5 : Update to 5.34.19

[cjones051073 (Chris Jones)]

bump to the latest version...

[mojca (Mojca Miklavec)]

Just a few questions before committing:

• Did you plan to include some code related to the (conflicting) python variants?
• We might want to address ticket #44224 (maybe by blacklisting compilers), but I'm not sure how and it's low priority anyway.
• Somewhere on the "TODO" list I have the following chunk of code left (probably it needs to be tested and compared with ROOT 6):

  +# make sure that the proper variant is automatically selected
  +# (ROOT crashes if the compiler it was built with doesn't exist)
  +if {![variant_isset clang33] && ![variant_isset clang34] && ![variant_isset clang35]} {
  +    if { ${configure.compiler} eq "macports-clang-3.3" } {
  +        default_variants-append +clang33
  +    } elseif { ${configure.compiler} eq "macports-clang-3.4" } {
  +        default_variants-append +clang34
  +    } elseif { ${configure.compiler} eq "macports-clang-3.5" } {
  +        default_variants-append +clang35
  +    }
  +}
  

[cjones051073 (Chris Jones)]

Replying to mojca@…:

Just a few questions before committing:

• Did you plan to include some code related to the (conflicting) python variants?

I don't think anything needs updating in the root5 port for this... It should not have the same issues as the root6 port has. ... Or has my memory missed something ?

• We might want to address ticket #44224 (maybe by blacklisting compilers), but I'm not sure how and it's low priority anyway.

I also do not know how...

• Somewhere on the "TODO" list I have the following chunk of code left (probably it needs to be tested and compared with ROOT 6):

  +# make sure that the proper variant is automatically selected
  +# (ROOT crashes if the compiler it was built with doesn't exist)
  +if {![variant_isset clang33] && ![variant_isset clang34] && ![variant_isset clang35]} {
  +    if { ${configure.compiler} eq "macports-clang-3.3" } {
  +        default_variants-append +clang33
  +    } elseif { ${configure.compiler} eq "macports-clang-3.4" } {
  +        default_variants-append +clang34
  +    } elseif { ${configure.compiler} eq "macports-clang-3.5" } {
  +        default_variants-append +clang35
  +    }
  +}
  

Yeah, I guess this could be included. I confess I did forget this. Please feel free to add it if you like... Its less important I think as root5 does not use any macports clang compiler by default, due to blacklists etc. So I think in practise hardly any uses will use it, and then probably via the variant anyway.

Chris

[cjones051073 (Chris Jones)]

Updated patch

[cjones051073 (Chris Jones)]

Patch file updated. Added the clang stuff + a few new features I noticed from the release notes...

[mojca (Mojca Miklavec)]

I'm not saying that this is a showstopper, but it's something I expected to fail based on problems in root6:

> port info root5 +python27
Error: Cannot install root5 with +python27 variant
Error: Unable to open port: root6 port is already installed with +python27 and both cannot be active at once. Pick a different python variant.

[mojca (Mojca Miklavec)]

I think that something like this (inside root5) should work, but I didn't test yet until pre-activate (and it doesn't prevent building root5 +python27 while having root6 +python27 installed):

    require_active_variants root6 "" python${ver_no_dot}

Maybe that's because root6 isn't a dependency of root5? This is certainly something that needs to be fixed though. (Either in my experimental code or in the portgroup.)

[cjones051073 (Chris Jones)]

Replying to mojca@…:

I'm not saying that this is a showstopper, but it's something I expected to fail based on problems in root6:

> port info root5 +python27
Error: Cannot install root5 with +python27 variant
Error: Unable to open port: root6 port is already installed with +python27 and both cannot be active at once. Pick a different python variant.

OK... I didn't think that was something anyone would try... I thought it more likely

 > port info root5
 > port variants root5

which still work...

If you really think it is needed, the same change could be made to the python variants I suppose, to move the check to a later phase...

Chris

[mojca (Mojca Miklavec)]

Also: I'm not saying that this is something of high importance. I can commit root5 without that change and leave dealing with python variants for later. In particular because we don't know yet how to fix this properly.

[cjones051073 (Chris Jones)]

Please commit it if you agree. Any changes to the python stuff should probably happen with the root6 port.

[mojca (Mojca Miklavec)]

Committed in r121910 with no changes to python.

[cjones051073 (Chris Jones)]

It appears upstream have decided to perform a stealth update on the 5.34.19 tarball, as I have just started to see checksum errors on a few machines. No idea why, as I cannot see any obvious reason from the git commits...

anyway, the patch coming next works around this in the usual way ...

Grrr...

Chris

[mojca (Mojca Miklavec)]

I knew I shouldn't have updated so fast ;)

[mojca (Mojca Miklavec)]

I would suggest the following patch instead:

Portfile

@@ -24 +24 @@
 master_sites        http://root.cern.ch/download/ \
                     ftp://root.cern.ch/root/
 
-checksums           rmd160  8e8a92d8734bd3e30abae66794dac07085e9b7dc \
-                    sha256  c516d167434c7d6d9c6a2f44f7b4af4115d3d0e3389ba79999ae984aee61356d
+checksums           rmd160  70dfc1db7198a34edf4a1a06f1750123852093ce \
+                    sha256  4d83310f2f22fed231220eabfccadcd335f8405d245ba136ea24a267efdcf5c6
 
+dist_subdir         root
 worksrcdir          root
 
 depends_lib         port:xz \

[cjones051073 (Chris Jones)]

Why ? I just followed the guidelines at

​https://trac.macports.org/wiki/PortfileRecipes

I'm fine either way, as long as it works. Just curious as to why the change...

Chris

[mojca (Mojca Miklavec)]

One reason to do it in a different way than the FAQ suggests is because we used to put all files under

/opt/local/var/macports/distfiles/root/

At some point (probably with version 5.34.38), but I'm not sure when we apparently (accidentally?) switched to

/opt/local/var/macports/distfiles/root5/

I see no reason to continue storing files under root5, we should rather switch back to root. And if we do that, we don't have to worry about stealth updates.

By comparing the two zips it looks as if windows project files had LF line endings. (I didn't inspect closely, that was just my first impression.) This is why we shouldn't need the revbump.

[cjones051073 (Chris Jones)]

Replying to mojca@…:

One reason to do it in a different way than the FAQ suggests is because we used to put all files under

/opt/local/var/macports/distfiles/root/

At some point (probably with version 5.34.38), but I'm not sure when we apparently (accidentally?) switched to

/opt/local/var/macports/distfiles/root5/

I see no reason to continue storing files under root5, we should rather switch back to root. And if we do that, we don't have to worry about stealth updates.

By comparing the two zips it looks as if windows project files had LF line endings. (I didn't inspect closely, that was just my first impression.) This is why we shouldn't need the revbump.

OK by me, makes some sense...

Chris

[mojca (Mojca Miklavec)]

Committed in r121937.