10:16 <GitHub111> [migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/6bd8566c48cd8a7d97541cd22d476e80fc1894e6

10:16 <GitHub111> [migen/master] setup.py: fix to catch all modules - Alain Péteut

19:30 <wpwrak> nice amount of spam on the blog :) (all caught by moderation, of course)

22:58 <lekernel> phew... yeah

22:58 <lekernel> the wiki was even worse until I switched it to manual account creation

22:58 <wpwrak> isn't it amazing how many parasites think our ecosystem can support them ? :)

22:59 <lekernel> also, mediawiki is a security disaster

23:00 <lekernel> spam bots happily hijacked the complete site layout - I couldn't even click the links to delete the pages

23:00 <lekernel> had to remove the junk directly from the DB

23:00 <wpwrak> nice ;-)

23:00 <wpwrak> makes one wonder how wikipedia survives

23:00 <lekernel> yeah I'm asking myself the same question

23:01 <lekernel> mediawiki is _ridiculous_

23:01 <lekernel> you can't mass delete content or users

23:01 <lekernel> you have to delete each vandalized page one by one, click each confirm link, etc

23:01 <lekernel> and you can't even delete a user account, only block it

23:02 <lekernel> for some reason, it seems that for the mediawiki developers, a feature 'delete a user account and all of its edits' is too much

23:03 <wpwrak> maybe they prefer to give their vermin that special touch of personal attention :)

23:04 <wolfspra1l> the mediawiki developers take considerable extra efforts to make sure that the mediawiki software works for tiny wikis like qi or milkymist at all, when their real focus and paycheck comes from making it work for wikipedia.org and other super high volume sites

23:04 <wolfspra1l> that's a big feature they are carrying around to support the small guys

23:04 <wolfspra1l> if they would only work on what *they* need, it would be un-installable for us in the first place...

23:05 <wolfspra1l> bottom line I cannot complain, it works

23:05 <lekernel> but it wastes your time compared to other wiki engines which deal with spam in a non-ridiculous manner

23:05 <wolfspra1l> then switch

23:06 <wpwrak> a git-based wiki would be sweet :) with an offline renderer

23:06 <wpwrak> ideally to ASCII :)

23:06 <lekernel> yeah, when I have enough users that taking time to convert the DB is justified

23:07 <lekernel> right now only me can create user accounts, and given the small size of the community, it solves the spam problem while wasting a relatively reasonable amount of my time

23:08 <wpwrak> lekernel: we should now be able to chase the strange patch compiler issues you're experiencing after my changes. with the patches i just posted in place, the regression tests should run the full set of patches with code generation enabled. anything as severe as the problems you saw should trip this.

23:09 <wpwrak> (unless, of course, it's a problem that only exists when building for lm32. that would suck ...)

23:30 <GitHub25> [milkymist] sbourdeauducq pushed 2 new commits to master: https://github.com/milkymist/milkymist/compare/14ad8ba...2ce4328

23:30 <GitHub25> [milkymist/master] software/libfpvm/Makefile: update dependencies - Werner Almesberger

23:30 <GitHub25> [milkymist/master] fpvm.c: suppress compiler warning caused by -Wunused-value - Werner Almesberger

23:33 <lekernel> wpwrak: for the default MMDIR, you can try something relative to the current path instead of /opt

23:34 <lekernel> and you have a .git extension when you clone ...?

23:36 <lekernel> from my understanding of git, things with a .git extension are internal git databases, not working directories

23:37 <wpwrak> i'm curious about that myself :) the stuff under /opt/ has that extension for some reason

23:38 <lekernel> from xiangfu's scripts?

23:39 <wpwrak> could be ... searching ...

23:39 <lekernel> https://github.com/milkymist/scripts/blob/master/compile-flickernoise/Makefile#L187

23:40 <lekernel> there should be no .git in the folder name

23:40 <lekernel> it's only because this script adds it ...

23:40 <wpwrak> yup. that's what i used.

23:40 <wpwrak> the "official" process :-)

23:42 <lekernel> I'm actually not using it ... had to build everything manually to get it working in the first place...

23:43 <lekernel> that might explain divergences between our binaries, like the compiler problem

23:43 <lekernel> in case some library causes subtle memory corruption

23:44 <wpwrak> cross-repository issues are always the most fun ...

23:44 <GitHub152> [flickernoise] sbourdeauducq pushed 5 new commits to master: http://git.io/SGrEVA

23:44 <GitHub152> [flickernoise/master] compiler/ptest: new option -c to run code generation - Werner Almesberger

23:44 <GitHub152> [flickernoise/master] compiler/test: test code generation and enable code generation for patch pool - Werner Almesberger

23:44 <GitHub152> [flickernoise/master] compiler/file2h: produce valid empty string if the file is empty - Werner Almesberger

23:46 <wpwrak> so you're saying nothing should use /opt/milkymist/ ?

23:46 <wpwrak> i'd certainly welcome getting rid of such a "magic" location. we still have /opt/rtems-* as a hard-coded shared place, but that one may be hard to get rid of

23:48 <lekernel> yeah, relative paths are a bit better

23:48 <wpwrak> infinitely better :)

23:48 <wpwrak> also because they allow you to have multiple concurrent versions. without having to hop around branches

23:57 <wpwrak> patch sent :)