version up : gcc 5.4.0.

component/gcc/buildout.cfg

@@ -44,8 +44,8 @@ environment =
 
 [gcc-common]
 recipe = slapos.recipe.cmmi
-url = http://ftp.gnu.org/gnu/gcc/gcc-5.3.0/gcc-5.3.0.tar.bz2
-md5sum = c9616fd448f980259c31de613e575719
+url = http://ftp.gnu.org/gnu/gcc/gcc-5.4.0/gcc-5.4.0.tar.bz2
+md5sum = 4c626ac2a83ef30dfb9260e6f59c2b30
 # make install does not work when several core are used
 make-targets = install -j1
 

[Vincent Pelletier]

@kazuhiko: @tc reports that this commit breaks cloudooo tests.

For a bit more about this commit: This is done because gcc 5.3.0 fails to build with gcc 6 (oh the irony).

@tc: What is the exact build issue this causes ? I could not access test logs (404). Also, I could locall build ERP5 SR, which semms to still include cloudooo as of current commit.

[Tristan Cavelier]

This is what I don't understand.. there's no error, but there is a strange log file... The file ends with a lot null characters \x00. (see suite.log https://lab.nexedi.com/snippets/155/raw 16.8MB) (also sent by mail)

[Kazuhiko Shiozaki]

Also, infinite fail results come from 'Destination path ......./fonts/ocrb-fonts/m4/m4 already exists'. It is unexpected because the configuration is like :

recipe = hexagonit.recipe.download
ignore-existing = true

[Vincent Pelletier]

"lots of null" smells like filesystem corruption (partial append). I would suspect the test node (or just that partition ?) to be in a weird state.