Hi, I think it would improve the quality of Haskell in Debian if we are more focused on packages that are relevant, and regularly clear out cruft. With cruft, I mean packages that were once dependencies of something we want to have in Debian, and are no longer, and also packages that someone used to be interested in and is no longer. So I propose to introduce “key packages”: These are those packages that should be in Debian on their own right – things like pandoc, xmonad, cabal-debian, etc. Packages become key packages if someone marks them as such in the package plan. Then, the following packages should be in Debian: * Key packages, * their dependencies, * and all packages that were previously in Debian and are also part of stackage (as I expect that these incur very little extra overhead). This implies that we should remove all packages that are neither in Stackage nor that are (used by) key packages. I added support for such reasoning to the package plan, and marked a few packages as key packages. This would currently imply that we can remove these packages: * ALUT * AMI * EdisonAPI * EdisonCore * GLFW * HCard * HDBC * HDBC-odbc * HDBC-postgresql * HDBC-sqlite3 * HFuse * HGL * HJScript * HJavaScript * HSH * HaskellForMaths * HsSyck * LDAP * LibZip * OSM * OpenAL * Ranged-sets * SDL * SDL-gfx * SDL-image * SDL-mixer * SDL-ttf * SMTPClient * Shellac * SimpleEA * Tensor * acid-state * algebra * augeas * bencode * binary-shared * bindings-gpgme * bindings-libzip * bindings-nettle * bindings-sane * bitarray * boomerang * bytestring-nums * bytestring-show * carettah * categories * cautious-file * cgi * chell-hunit * citeproc-hs * clocked * cmark * config-value * crypto-conduit * crypto-pubkey-openssh * cryptonite * data-accessor-template * data-lens * data-lens-template * data-pprint * datetime * diagrams-gtk * download-curl * dpkg * ed25519 * event-list * explicit-exception * failure * ftphs * gconf * github * graceful * gtksourceview2 * gtksourceview3 * haddock * happstack * happstack-authenticate * happstack-heist * happstack-hsp * harp * hashed-storage * haskell-qrencode * haskelldb * haskelldb-hdbc * haskelldb-hdbc-odbc * haskelldb-hdbc-postgresql * haskelldb-hdbc-sqlite3 * haskore * hbro * hbro-contrib * hcwiid * hedis * hothasktags * hscurses * hsmagick * hsp * hsql * hsql-mysql * hsql-odbc * hsql-postgresql * hsql-sqlite3 * hsqml * hsx-jmacro * hsx2hs * hxt-cache * hxt-xpath * hxt-xslt * iCalendar * intern * io-streams * iteratee * knob * language-python * largeword * leksah * leksah-server * libsystemd-journal * ltk * maccatcher * magic * markov-chain * mersenne-random * midi * mime * monoid-transformer * murmur-hash * ncurses * netwire * non-negative * nonce * numeric-quest * ofx * openpgp * pipes-zlib * ranges * reactive-banana * readline * reform * reform-blaze * reform-hamlet * reform-happstack * reform-hsp * regex-pcre * regex-tdfa-text * regex-tdfa-utf8 * rosezipper * sandi * sfml-audio * simpleirc * snap-loader-dynamic * snap-loader-static * snaplet-acid-state * strptime * svgcairo * swish * syb-with-class-instances-text * tasty-hspec * template * test-framework-th-prime * torrent * universe-base * unix-bytestring * utility-ht * vado * vcsgui * vcswrapper * vte * wai-handler-fastcgi * web-plugins * web-routes * web-routes-boomerang * web-routes-happstack * web-routes-hsp * web-routes-th * webkit * x509-util * xcb-types * xml2html * yaml-light * yesod-auth-account * yesod-default * yesod-markdown * yesod-routes Quite a list! Do you spot anything that you want to keep? Then mark it as "key" in the package list (but be prepared that I come to you when this package needs extra work) and/or make sure it becomes part of stackage. Do you think this system is sensible? Will it scale? Greetings, Joachim -- Joachim "nomeata" Breitner Debian Developer nomeata@debian.org | ICQ# 74513189 | GPG-Keyid: F0FBF51F JID: nomeata@joachim-breitner.de | http://people.debian.org/~nomeata
Attachment:
signature.asc
Description: This is a digitally signed message part