Bugzilla Bug 115468

I would like all supporting arches to please stabilise    
sci-biology/phylip-3.65. Here is how to test the package:    
     
Prepare the testing environment:      
  $ mkdir phylip-test      
  $ cd phylip-test      
  $ wget pelican.rsvs.ulaval.ca/distfiles/seq/glxrs.phylip     
  $ wget pelican.rsvs.ulaval.ca/distfiles/seq/consense-ref.tree     
  $ cp /usr/share/phylip/fonts/font1 .     
     
Bootstrap the alignment (create 100 small sequence variations that will be     
used to assess the robustness of the tree):     
  $ cp glxrs.phylip infile      
  $ seqboot      
  (Accept default parameters, use seed 13.)      
  $ cp outfile seqboot.out      
  $ rm infile outfile     
     
Construct phylogenetic trees (one for each of the 100 variants) using the     
parsimony algorithm:     
  $ cp seqboot.out infile      
  $ protpars     
  (Accept default parameters.)     
  $ cp outfile protpars.out     
  $ cp outtree protpars.tree     
  $ rm infile outfile outtree     
     
Calculate the consensus for all 100 trees:     
  $ cp protpars.tree intree     
  $ consense     
  (Accept default parameters.)     
  $ cp outfile consense.out     
  $ cp outtree consense.tree     
  $ rm intree outfile outtree     
     
Compare the tree with the reference one:     
  $ diff -u consense-ref.tree consense.tree     
  (There should not be any difference between the two files.)     
     
Draw the tree on screen:     
  $ cp consense.tree intree     
  $ drawgram     
  (Use font1 as the fontfile, accept default parameters.)     
  (The tree should be drawn in a new X window.)     
  File > Quit   
   
If the computed tree is identical to the reference one, and the tree can be   
drawn on screen, PHYLIP is working.  
 
Thanks in advance,

This is actually a blocker.  THis bug cannot go to stable before this collision
is fixed.

Everything works according to test procedure.  Recommend stable after collision
fix.

Collision fixed. Ebuild to stabilise is 3.65-r1. 

DEPEND="virtual/libc
        virtual/x11"

phylip is not part of the system and as such the virtual/libc dependency should
not be there.

I am unclear about the dependency on virtual/libc. It has been in the ebuild   
ever since it was commited to Portage and there are lots of other packages   
that depend on libc that have nothing to do with system. On the other hand,  
most packages in the tree depend on libc but do not declare it. Could you  
explain what virtual/libc is for in more details? Should dependencies on  
virtual/libc be removed from all non-system packages? (I could not find a 
tracker bug for that.) 

(In reply to comment #5)
> I am unclear about the dependency on virtual/libc. It has been in the ebuild   
> ever since it was commited to Portage and there are lots of other packages   
> that depend on libc that have nothing to do with system. On the other hand,  
> most packages in the tree depend on libc but do not declare it. Could you  
> explain what virtual/libc is for in more details? Should dependencies on  
> virtual/libc be removed from all non-system packages? (I could not find a 
> tracker bug for that.) 

http://dev.gentoo.org/~plasmaroo/devmanual//general-concepts/dependencies/

Yes, not everything in the tree is up to standards so that is why I like to do
checkups when people are asking stuff to be marked stable.

virtual/libc exists because we provide different alternatives for a c library
like uclibc.

Thanks for the explanation, Petteri. I removed the dependency (and will do so 
for other packages the sci herd maintains). 

(In reply to comment #0)
>   $ drawgram     
>   (Use font1 as the fontfile, accept default parameters.)     
>   (The tree should be drawn in a new X window.)     
>   File > Quit   
>    
> If the computed tree is identical to the reference one, and the tree can be   
> drawn on screen, PHYLIP is working.

This is the only part that fails on ppc-macos on my end. Otherwise, things work great. I'm not sure that 
phylip's X stuff is even supposed to work on Apple's X11 (there are native packages for the phylip GUI 
stuff IIRC). Apart from that, I'm happy with it. Let me know what you think.

Hasan:  
  
Your call. However, since there is already a stable version, I think you  
should either stabilise this one or put 3.64 back to ~ppc-macos too. If you  
decide to stabilise, maybe you should add a note regarding X11 support for  
MacOS users in the README.Gentoo.  

also read the ppc-macos keywording policy thing, please.

stable on x86, thanks for the test cases :)

tested on ppc
works for me

ppc stable, thanks for testing nixnut :)

Sorry about the delays --  have been travelling.

I've dropped the stable keyword from phylip-3.64, in accordance with my
interpretation of the intentions of the ppc-macos interim keywording policy.

Cannot be stabilised on ppc-macos.