| 1 |
Mark Cox wrote: |
| 2 |
> On Tue, 2002-07-09 at 00:25, Chad M. Huneycutt wrote: |
| 3 |
> |
| 4 |
>>Mark Cox wrote: |
| 5 |
>> |
| 6 |
>>>opencm (www.opencm.org) wants to generate ps documents from info files |
| 7 |
>>>when you tell it to compile. The problem is sandbox (I think it is this |
| 8 |
>>>that denies access to everything outside of the work dir..someone |
| 9 |
>>>correct if it isn't.) causes this to happen . |
| 10 |
>>> |
| 11 |
>>> |
| 12 |
>>>dvips -t letter opencm.dvi -o tmp.ps |
| 13 |
>>>This is dvips(k) 5.86 Copyright 1999 Radical Eye Software |
| 14 |
>>>(www.radicaleye.com) |
| 15 |
>>>' TeX output 2002.07.08:1220' -> tmp.ps |
| 16 |
>>>kpathsea: Running mktexpk --mfmode ljfour --bdpi 600 --mag 1+437/600 |
| 17 |
>>>--dpi 1037 cmbx12 |
| 18 |
>>>ACCESS DENIED mkdir: /usr/share/texmf/fonts/pk |
| 19 |
>>>mkdir: cannot create directory `/usr/share/texmf/fonts/pk': Permission |
| 20 |
>>>denied |
| 21 |
>> |
| 22 |
>> |
| 23 |
>>Is this an ebuild you are testing, or is this one in portage? The first |
| 24 |
>>line up there needs to be removed from the install process. It might |
| 25 |
>>possibly be moved to another section of the ebuild, or just removed |
| 26 |
>>completely. If this is in portage, then if no bug exists on |
| 27 |
>>bugs.gentoo.org, please submit one. If you really want to install it over |
| 28 |
>>the sandbox violation, then do |
| 29 |
>> |
| 30 |
>># FEATURES="-sandbox" ebuild ... |
| 31 |
> |
| 32 |
> |
| 33 |
> Sorry I always I leave something important out.... Yes this is an ebuild |
| 34 |
> I'm testing. And no it isn't in the portage repository either. |
| 35 |
> |
| 36 |
> I like the idea of sandbox as I don't trust the program (or any program) |
| 37 |
> to be 100% correct (especially in an alpha release). |
| 38 |
> |
| 39 |
> Is this a bug in portage I don't think so. Maybe sandbox may have to |
| 40 |
> make exceptions for TeX, or is there a way for TeX to all of these |
| 41 |
> metafonts (if thats what its trying to create ??) when you emerge TeX. |
| 42 |
|
| 43 |
One solution would be to remove this from the install process and put it in |
| 44 |
the config step. The config step happens after merge, so the sandbox |
| 45 |
checking is no longer in effect. |
| 46 |
|
| 47 |
You might also check the tetex ebuild and see what kind of tricks they do |
| 48 |
there. |
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
-- |
| 53 |
Chad Huneycutt (chadh@g.o) |
Archives