#2427: Allow compilation of source from stdin
-----------------------------+----------------------------------------------
Reporter: guest | Owner:
Type: feature request | Status: new
Priority: normal | Milestone:
Component: Compiler | Version: 6.8.3
Severity: minor | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: Unknown
Os: Unknown |
-----------------------------+----------------------------------------------
Changes (by igloo):
* difficulty: => Unknown
Old description:
> Hiya. So, as part of how the Hint library (a wrapper around the GHC API
> operates), it has to generate a file in /tmp containing module
> boilerplate and the supplied code of interesting, and it then evaluates
> it*. But for my mueval code (which uses Hint), I'd like to disable file
> creation entirely through resource limits, and so it would be much better
> for Hint if it could instead just create the string and pipe it right
> into GHC. This avoids any file creation (which may not be possible for
> any number of reasons besides resource limits, like LiveCDs or read-only
> disks). But in my experiments, and those of #haskell, GHC determinedly
> blocks any attempt - you can't simply pipe it in with |, you can't use
> /dev/stdin, can't use one of the file descriptors, etc.
>
> So what I would like is to be able to do:
> $ echo "import qualified Data.List\nmain = print $ Data.List.intersperse
> 'f' "ggggg"
>
> "fgfgfgfgfgfg"
>
> (Why can't I use ghc -e? Doesn't do imports of any kind, and if mueval is
> to replicate lambdabot's functionality, it needs to be able to import
> many libraries qualified.**)
>
> * It has to do this roundabout hackish thing because alas, the GHC API
> seems to allow functions from any module whatsoever to be called as long
> as they are named qualified, even if the appropriate module had not been
> allowed in. The magnitude of this as a security hole is obvious.
>
> ** Why qualified? Too many of the basic libraries have name collisions.
> Can't remove them, as it'd be silly to have only half the base libraries
> or whatever, but can't just blindly import them as you'll get conflicts.
New description:
Hiya. So, as part of how the Hint library (a wrapper around the GHC API
operates), it has to generate a file in /tmp containing module boilerplate
and the supplied code of interesting, and it then evaluates it*. But for
my mueval code (which uses Hint), I'd like to disable file creation
entirely through resource limits, and so it would be much better for Hint
if it could instead just create the string and pipe it right into GHC.
This avoids any file creation (which may not be possible for any number of
reasons besides resource limits, like LiveCDs or read-only disks). But in
my experiments, and those of #haskell, GHC determinedly blocks any attempt
- you can't simply pipe it in with |, you can't use /dev/stdin, can't use
one of the file descriptors, etc.
So what I would like is to be able to do:
{{{
$ echo "import qualified Data.List\nmain = print $ Data.List.intersperse
'f' "ggggg"
"fgfgfgfgfgfg"
}}}
(Why can't I use ghc -e? Doesn't do imports of any kind, and if mueval is
to replicate lambdabot's functionality, it needs to be able to import many
libraries qualified.**)
* It has to do this roundabout hackish thing because alas, the GHC API
seems to allow functions from any module whatsoever to be called as long
as they are named qualified, even if the appropriate module had not been
allowed in. The magnitude of this as a security hole is obvious.
** Why qualified? Too many of the basic libraries have name collisions.
Can't remove them, as it'd be silly to have only half the base libraries
or whatever, but can't just blindly import them as you'll get conflicts.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2427#comment:3>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
