[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Lambdabot?



On Sat, 22 Aug 2015 09:15:36 +0200
Alexandre Delanoë <anoe@debian.org> wrote:

> I pushed on anoe branch. If someone could check this, thank you in
> advance.

I've checked your branch. Here is what I've noticed by package):

haskell-lambdabot:
  There is already a package lambdabot (without the haskell- prefix).
  You should upgrade that, instead of packaging it with another name.

haskell-lambdabot-core:
  The control files contains these Build-Dependencies[-Indep]:
    Build-Depends:
    [...]
    libghc-parsec3-dev (>= 3),
    libghc-parsec3-prof (>= 3) | libghc-parsec2-prof (<< 3),
    [...]
    Build-Depends-Indep:
    [...]
    libghc-parsec3-doc (>= 3) | libghc-parsec2-doc (<< 3),
  This doesn't look right. Why are the alternatives only for -prof and
  -doc packages? I would suggest dropping the alternative
  build-dependencies on libghc-parsec2-* completely.

haskell-lambdabot-trusted:
  The description contains the sequence "\@". The backslash is there to
  prevent hackage from parsing the "@" as markup. We don't have this
  markup in our package descriptions, so the backslash needs to go away.

haskell-misfortune:
  This package depends on the -dev packages of monad-loops, regex-base
  and reges-pcre but not on the corresponding -prof and -doc packages.
  This doesn't seem right, because we always build the -prof and -doc
  packages but if there is some reason for this, just tell me.

  cabal-debian generated an executable packages with the standard name
  haskell-misfortune-utils. This name isn't very nice and we usually
  pick a better name for the package. Maybe something like "misfortune"
  would suite the package better.

Building stuff against experimental:
  Because we are currently transitioning towards GHC 7.10, we need to
  ensure that the buildds build our packages against the version from
  experimental, however, they prefer to build against packages from
  unstable. We need to force them to build against the right version,
  by making all packages depend on haskell-devscripts >= 0.10 (version
  depending on current transition), which pulls in the right GHC.
  Normally this is done all at once but Joachim already did this for
  all other packages on the experimental branch, so please add that
  dependency to you packages, before your changes get merged there.

Some general problems with copyright:
  Most copyright files have serious problems, like missing some
  copyright holders listed in source files or not containing the right
  license text.
  
  Some copyright holders listed in individual source files aren't
  listed in the .cabal file and cabal-debian didn't write them in the
  resulting copyright file.
  "find -name '*.hs' -exec head '{}' +" (run from the package directory
  with unpacked upstream sources) gives you the first few lines of
  every Haskell file in the project. You can go through this list to
  find files with copyright holders different from those listed in
  the .cabal file.

  When the .cabal file says the license is GPL it actually means GPL-2.
  Debian copyright files distinguish this. Changing GPL to GPL-2 in the
  copyright files, should be sufficient.

  cabal-debian seems to have placed some of the license texts in the
  comment section of the license paragraph. (afaik) This is wrong.
  Pleas remove the Comment: part on the first line of the license text
  in the affected files (e.g. haskell-lambdabot-core's copyright file)
  and align the first line with the following lines.

  The upstream LICENSE.txt files fail don#t contain the exact terms of
  the GPL-2 and neither state were they can be found. However, Debian
  policy requires this. Adding something like "The exact terms and
  conditions of the GNU General Public license Version 2 can be found
  in /usr/share/common-licenses/GPL-2." to the license text, should
  fix this.

  Some copyright files were generated with an empty line between the
  first "Copyright:" field and the corresponding "License:" field. This
  is invalid, because they're seen as different paragraphs this way.

I'm afraid I was a bit unclear on some points (especially the
copyright stuff). So if you need clarification or help on some points,
just ask.

Regards
Sven


Reply to: