On 2014-12-12, kcrisman <
kcri...@gmail.com> wrote:
> ------=_Part_154_1466710874.1418353858850
> Content-Type: multipart/alternative;
> boundary="----=_Part_155_44875405.1418353858850"
>
> ------=_Part_155_44875405.1418353858850
> Content-Type: text/plain; charset=UTF-8
>
>
>
>> Sage (6.4.1) notebook build failed on 32 bit CentOS 5.11. Version of
>> sagenb was 0.11.1. To reproduce, build Sage from source (e.g., ./make from
>> Sage installation directory). CPU was "CPU1: Intel(R) Core(TM)2 Duo CPU
>> E4400 @ 2.00GHz stepping 0d" per dmesg. Sage runs, but ptest or building
>> pdf docs replays the error messages, as shown below. The tar file for
>> sagenb could be unpacked manually OK (in off-tree location). Checksums on
>> Sage source download tarball matched up OK. What is the best course of
>> action to get this to build properly? Is there a reason why it failed?
>> I previously posted a similar item, but it did not appear in the
>> sage-support list of items, so I am trying it again.
>>
>>
> I wonder if it's because of the bsd tar I used to tar it (I built that
don't you have 'gtar' on your machine too?
surely enough, I see the same warnings on a 64-bit Linux:
$ tar tf upstream/sagenb-0.11.1.tar | less
tar: Ignoring unknown extended header keyword `LIBARCHIVE.creationtime'
tar: Ignoring unknown extended header keyword `SCHILY.dev'
tar: Ignoring unknown extended header keyword `SCHILY.ino'
....
Probably my tar is newer and thus the install can live with this.
IMHO this tarball should be replaced by something glatt kosher...
Please open a ticket and cc me on it.
>
> What you could do is to retar the directory yourself, and replace the old
> tar file with that one; then things should work fine.
no, this is not enough, as this will surely change the tarball
checksum; you need to follow the instructions
in sage developer guide to recreate them.
http://sagemath.org/doc/developer/packaging.html#checksums
Best,
Dima