Hi Kiwamu, Am Samstag, den 22.09.2012, 15:29 +0900 schrieb Kiwamu Okabe: > I think newer /usr/sbin/update-hoogle is too slow for old PC. > > root@casper:~# echo 3 > /proc/sys/vm/drop_caches > root@casper:~# time find /usr/share/doc -wholename > "/usr/share/doc/ghc-doc/html/libraries/*/*.txt*" -or -wholename > "/usr/share/doc/libghc-*-doc/html/*.txt" > /dev/null > > real 3m15.971s > user 0m0.380s > sys 0m1.300s > root@casper:~# echo 3 > /proc/sys/vm/drop_caches > root@casper:~# time find /usr/share/doc/ghc-doc/html/libraries/ -name > "*.txt*" > /dev/null; time find /usr/share/doc/libghc-*-doc/ > -wholename "/usr/share/doc/libghc-*-doc/html/*.txt" > /dev/null > > real 0m0.802s > user 0m0.000s > sys 0m0.020s > > real 0m3.512s > user 0m0.040s > sys 0m0.080s > > Can I roll back update-hoogle to old version? sorry for the regression, certainly we can do better. I do not like this find-based approach at all. After all, what happens if some package ships an "example.txt" in their documentation – wouldn’t things break? Looking through your code again, I am still confused why you need to such things at install time at all. Can’t you, at build time of the package in question (i.e. in ghc and via haskell-devscripts) already fix the .txt file (i.e. the effect of haddoc-collect) and ship the resulting file in, say, /usr/lib/ghc-doc/hoogle/ (which would nicely sit next to /usr/lib/ghc-doc/haddock/). As a side effect this avoids the compression by dh_comporess. Then the hoogle trigger only needs to watch precisely this directory, and hoogle-update only needs to convert the files there and run "hoogle combine" on the result. Also, is there any reason to keep the .hoo files (besides default.hoo) around? Greetings, Joachim -- Joachim "nomeata" Breitner Debian Developer nomeata@debian.org | ICQ# 74513189 | GPG-Keyid: 4743206C JID: nomeata@joachim-breitner.de | http://people.debian.org/~nomeata
Attachment:
signature.asc
Description: This is a digitally signed message part