I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable. I didn't offer an opinion on the matter. I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason. I'd appreciate some comments on this issue. Thanks. -- Tony Morris http://tmorris.net/
Tony Morris
I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable.
Huh? That's pretty weird.
I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason.
Well - one common setup is to create users along with their own group, and optional membership in other groups, combined with an umask of 002 (i.e., files are created group-writeable) or perhaps 027 (group-readable, but not accessible at all to others). This way, a group-shared directory can have the setgid bit set, and when a user writes a file in this directory, the file will inherit the group from the directory, and be writeable by other people of the group. Unless it also has the sticky bit set. (Hmm, aren't Unix permissions supposed to be the *simple* solution? It's been a while since I looked at this, so please correct if I got the details wrong. :-) -k -- If I haven't seen further, it is by standing in the footprints of giants
Am 09.12.2010 08:01, schrieb Tony Morris:
I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable. I didn't offer an opinion on the matter. I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason.
I'd appreciate some comments on this issue. Thanks.
Indeed I get:
GHCi, version 7.0.1: http://www.haskell.org/ghc/ :? for help Loading package ghc-prim ... linking ... done. Loading package integer-gmp ... linking ... done. Loading package base ... linking ... done. Loading package ffi-1.0 ... linking ... done. *** WARNING: /home/maeder/.ghci is writable by someone else, IGNORING!
and I'm surprised! Cheers Christian
On 09/12/10 07:01, Tony Morris wrote:
I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable. I didn't offer an opinion on the matter. I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason.
I'd appreciate some comments on this issue. Thanks.
You can insert *any* commands in the ghci file. including some system commands, to delete everything. That's probably the reason to deny a file to be group writable (anyone in a group could write bad thing in there) you can try putting the following in the .ghci to check for example: <<< readFile "somefileyouwanttoread" >>= writeFile "/tmp/apublicplace"
-- Vincent
Perhaps ghc should also ignore all group-writable *.hs, *.lhs, *.c, *.o, *.hi files.
On 10-12-09 01:57 PM, Claus Reinke wrote:
Perhaps ghc should also ignore all group-writable *.hs, *.lhs, *.c, *.o, *.hi files.
dot-ghci files are *run* if you just start ghci (or ghc -e) in that directory (even if you don't intend to compile, load, or run any Haskell code).
Haskell developers don't just run ghci. They go ahead to run arbitrary Haskell code too.
On Thu, 2010-12-09 at 14:01 -0500, Albert Y. C. Lai wrote:
On 10-12-09 01:57 PM, Claus Reinke wrote:
Perhaps ghc should also ignore all group-writable *.hs, *.lhs, *.c, *.o, *.hi files.
dot-ghci files are *run* if you just start ghci (or ghc -e) in that directory (even if you don't intend to compile, load, or run any Haskell code).
Haskell developers don't just run ghci. They go ahead to run arbitrary Haskell code too.
On the other hand - code that they likely just seen. On the other hand .ghci file may not have been even noticed. It makes it much lower risk. Regards
On Thu, 2010-12-09 at 17:01 +1000, Tony Morris wrote:
I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable. I didn't offer an opinion on the matter. I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason.
I'd appreciate some comments on this issue. Thanks.
While I may not understend setting it may be possible to set a repository in shared directory (like git, svn or darcs) and use non-group-writable copies to actual work. Regards
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 12/9/10 02:01 , Tony Morris wrote:
I teach haskell quite a lot. I recommend using .ghci files in projects. Today I received complaints about the fact that ghci will reject .ghci if it is group-writeable. I didn't offer an opinion on the matter. I am wondering if these complaints have legitimate grounds i.e. maybe you want to have group write on that file for some reason.
Linux likes to do the group-write thing (this derives from an early Debian decision that uids were obsolete and everything should use the gid as the uid; basically, the wrong solution to file ACLs). Given their decision and its near ubiquity in the Linux world these days, I suspect ghc's going to be on the wrong end of the battle. - -- brandon s. allbery [linux,solaris,freebsd,perl] allbery@kf8nh.com system administrator [openafs,heimdal,too many hats] allbery@ece.cmu.edu electrical and computer engineering, carnegie mellon university KF8NH -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.10 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk0EHvoACgkQIn7hlCsL25WM3ACeMMSJAcyBEM0KRkK71nhpCiOx pRMAn01M0F2AJlkm+IxIeMrnIral/vwS =DcFw -----END PGP SIGNATURE-----
participants (8)
-
Albert Y. C. Lai -
Brandon S Allbery KF8NH -
Christian Maeder -
Claus Reinke -
Ketil Malde -
Maciej Piechotka -
Tony Morris -
Vincent Hanquez