2013.4.0.0 proposal

[John Lato]

On Nov 11, 2013 8:43 AM, "Brandon Allbery" <allbery.b at gmail.com> wrote:
>
>
> On Mon, Nov 11, 2013 at 3:34 AM, Joachim Breitner <
mail at joachim-breitner.de> wrote:
>>
>> I’d feel more comfortable with this boot-library version deviation if
>> someone who knows the interaction of GHC (the compiler), ghc (the
>> librarY) and Cabal better can comment on it.
>
>
> I can't say I know the full details of the interactions, but the Cabal
library is essentially the only bootlib that not only can safely (I don't
think I've ever heard of anyone getting into trouble) be upgraded, but
often is *recommended* for upgrade to support newer cabal-install. I think
this is because ghc only uses a small portion of it that has been stable
for a long time (avoiding data incompatibility) and it's compiled in (so no
link time strangeness). The only issues that come to mind would be if
someone were to write TH that itself explicitly called out to Cabal, or
maybe some complex ghc-as-a-library stuff in a program explicitly using
Cabal itself.

TBH I don't see how even those cases would cause a problem. I think a more
likely situation would be a user installing a package that depends on the
boot-lib install, then have that package be incompatible with everything
else. But even that seems fairly rare.
John L.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/libraries/attachments/20131111/499175be/attachment.html>