Haskell › Haskell - Glasgow-haskell-bugs

[GHC] #5977: Allow ignoring global package db

Classic

List

Threaded

14 messages Options

GHC

[GHC] #5977: Allow ignoring global package db

#5977: Allow ignoring global package db
------------------------------+---------------------------------------------
Reporter: duncan | Owner:
Type: feature request | Status: new
Priority: normal | Component: Compiler
Version: 7.4.1 | Keywords:
Os: Unknown/Multiple | Architecture: Unknown/Multiple
Failure: None/Unknown | Testcase:
Blockedby: | Blocking:
Related: |
------------------------------+---------------------------------------------
For sandboxing, users want to be able to use only local package dbs. To
handle rts, base etc, they copy those specific packages from the global
package db into a local package db.

Currently however there is no sensible way to tell ghc to look '''only'''
at the listed package dbs, and not the global db. There is a flag `-no-
user-package-conf` to avoid looking at the per-user one, but no equivalent
to ignore or to specify the global one.

(Note: as a quirky non-sensible way, you can use `GHC_PACKAGE_PATH=""` to
clear the package db stack, but this should be possible using command line
flags.)

The user interface we're now using in cabal is like this:

* `--global` sets the initial db stack to `[GlobalPackageDB]`
* `--user` sets the initial db stack to `[GlobalPackageDB,
UserPackageDB]`
* `--package-db=clear` clears the db stack to `[]`
* `--package-db=global` pushes `GlobalPackageDB` on the top of the stack
* `--package-db=user` pushes `UserPackageDB` on the top of the stack
* `--package-db=`''${otherfile}'' pushes `SpecificPackageDB`
''${otherfile}'' on the top of the stack

It seems to me we could do something similar with ghc. Currently in ghc we
have a `-hide-all-packages` flag that resets the package set to empty. We
could do with an equivalent for the package database stack. I suggest
something like:

* `-clear-package-conf`
* `-global-package-conf`
* `-user-package-conf` (and the existing `-no-user-package-conf`)
* `-package-conf` file

The default would remain that the global and user dbs are used, but this
can be reset using `-clear-package-conf`.

Personally I prefer the nomenclarture "package db" rather than "package
conf" here, but it's not that important.

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner:
Type: feature request | Status: new
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------
Changes (by simonmar):

* priority: normal => high
* difficulty: => Unknown
* milestone: => 7.6.1

Comment:

Yep, let's do something for 7.6.1.

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner: pcapriotti
Type: feature request | Status: patch
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------
Changes (by pcapriotti):

* status: new => patch

Comment:

The attached patch series adds the extra flags described in this ticket
(with the addition of `no-global-package-conf`, for symmetry), and renames
them to `*-package-db`.

You can find the corresponding patch for Cabal in my github fork:
https://github.com/pcapriotti/cabal

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner: simonmar
Type: feature request | Status: patch
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------

Comment(by simonmar):

I had a look at the patches. The situation in `DynFlags` is now a bit
strange - we have both the `Opt_ReadGlobalPackageConf` options ''and'' the
`[PkgConfRef]`. I think we should have just the `[PkgConfRef]`
representing the stack, and all the flags should manipulate it directly.

So

* We start with `[global,user]`
* `-no-user-package-db` removes user from the list (similarly for global)
* `-user-package-db` adds user to the end (similarly for global)
* `-package-conf file` adds file to the end
* `-clear-package-db` sets the list to empty

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner: pcapriotti
Type: feature request | Status: patch
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------

Comment(by pcapriotti):

I'm not sure how that would interact with GHC_PACKAGE_PATH. For example,
if I run:

{{{
GHC_PACKAGE_PATH="foo:" ghc --no-user-package-db
}}}

I'd expect it to use `[global, foo]` as db stack, but we can't remove
`user` from the stack originating from `GHC_PACKAGE_PATH` unless we keep
the information somewhere (currently it's in `Opt_ReadUserPackageConf`).

So at the moment there is an "initial stack", which is `[global, user]` by
default, and is affected by `-no-user-package-db`, `-no-global-package-
db`, `-clear-package-db`, and `GHC_PACKAGE_PATH`. The other flags append
to the stack *after* the initial stack has been determined.

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner: pcapriotti
Type: feature request | Status: patch
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------

Comment(by simonmar):

I find it weird that `-no-user-package-db` affects the "initial stack",
but `-user-package-db` appends the user db to the final stack. So e.g.
`-user-package-db -no-user-package-db` does not do what you expect.

So I see two alternatives. One is to modify my proposal to take into
account `GHC_PACKAGE_PATH`: just initialise the stack from
`GHC_PACKAGE_PATH` before processing the arguments, defaulting to
`[global,user]` if `GHC_PACKAGE_PATH` is not set.

The other alternative is to keep the concept of the "initial stack", but
make `-user-package-db` and `-global-package-db` just the inverses of
`-no-user-package-db` and `-no-global-package-db` respectively. I'd be ok
with that too.

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:9>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
---------------------------------+------------------------------------------
Reporter: duncan | Owner: pcapriotti
Type: feature request | Status: patch
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Keywords: | Os: Unknown/Multiple
Architecture: Unknown/Multiple | Failure: None/Unknown
Difficulty: Unknown | Testcase:
Blockedby: | Blocking:
Related: |
---------------------------------+------------------------------------------

Comment(by p.capriotti@…):

commit 6a831be4aa73e86568256813ffa862d7cfd5732d
{{{
Author: Paolo Capriotti <[hidden email]>
Date: Thu May 3 11:29:51 2012 +0100

Add flags to manipulate package db stack (#5977)

Introduce new flags to allow any package database stack to be set up.
The `-no-user-package-conf` and `-no-global-package-conf` flags remove
the corresponding package db from the initial stack, while
`-user-package-conf` and `-global-package-conf` push it back on top of
the stack.

compiler/main/DynFlags.hs | 24 +++++++++++--
compiler/main/Packages.lhs | 81
+++++++++++++++++++++-----------------------
2 files changed, 60 insertions(+), 45 deletions(-)
}}}

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:12>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

GHC

Re: [GHC] #5977: Allow ignoring global package db

In reply to this post by GHC

#5977: Allow ignoring global package db
-------------------------------+--------------------------------------------
Reporter: duncan | Owner: pcapriotti
Type: feature request | Status: closed
Priority: high | Milestone: 7.6.1
Component: Compiler | Version: 7.4.1
Resolution: fixed | Keywords:
Os: Unknown/Multiple | Architecture: Unknown/Multiple
Failure: None/Unknown | Difficulty: Unknown
Testcase: | Blockedby:
Blocking: | Related:
-------------------------------+--------------------------------------------
Changes (by pcapriotti):

* status: patch => closed
* resolution: => fixed

--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/5977#comment:13>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

_______________________________________________
Glasgow-haskell-bugs mailing list
[hidden email]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs