No, that was not the fuzzy matching. That were the changes introduced to WS/1 when introducing NGS.
(Not WS/2)
WS/2 was fine, that was new atm.
I do understand it was difficult anyways to keep WS/1 working with NGS, though.
kepstin
getting individual disc lookups to work in ws/1 would have been very hard. would have had to maintain a mapping table of medium mbids following merges, and generate artificial mbids for each disc in native multi-disc releases
and the release-event splitting complicated things more, because there are suddenly multiple results for something when there used to be just one
JonnyJD
Yep.
Still much better than the option to deactivate WS/1.
ruaok joined the channel
ruaok
note to self. don't trip the powerstrip
warp
at least that is a power failure of your own doing
(in this country, it is definitely useful to have a laptop with working battery :)
ruaok
I bet.
warp
"only" one power outage since I've arrived, and that lasted 20 seconds.
Ian McEwen: mapping: support country, and display it
Sebastinas
JonnyJD: I guess cygwin should use the __declspec stuff
JonnyJD
Sebastinas: well, it doesn't work. But it possibly also gives no warning, similar to MinGW
Sebastinas
I can look into it.
But probabaly not tonight
kepstin
cygwin does some really strange stuff with linking; i've never been quite sure how that worked :(
i know that everything in cygwin is basically prelinked with static loading offsets and stuff, and when you start building your own libraries you might get address conflicts and the whole thing blows up
they have a tool to run that reassigns address spaces to fix that...
JonnyJD
all cygwin libraries are linked to special cygwin dlls. So they work on windows, but only if you add these cygwin dlls
kepstin
and the cygwin dlls are all gpl2, which makes licensing of binaries compiled with cygwin... interesting.
unless they changed that at some point
ah, "As a special exception to GPLv3+, Red Hat grants you permission to link software whose sources are distributed under a license that satisfies the Open Source Definition with libcygwin.a, without libcygwin.a itself causing the resulting program to be covered by GPLv3+."
ianmcorvidae
one of my friends who works at amazon says that apparently they're not allowed to use cygwin at all, may be related
Sebastinas: thanks. Feel free to push that to a branch somewhere (you don't necessarily have to open a request, I can also pull that manually). I'm having dinner now.
MBJenkins
Ian McEwen: import: import countries where they exist
Ian McEwen: display: show countries in search results
JonnyJD
ah, I didn't yet try MinGW on Linux. That explains the problem. But having that working would be nice.
Sebastinas: on automake/Windows libdiscid_EXPORTS seems broken. (cmake was the major build system on Windows before) Even fixing that (defining manually this doesn't seem to work)
On cmake I also can't get it to work somewhow. both, declspec and attribute work in general (no warnings), but with nm everything is still there.
Not sure if testing with "nm" makes any sense for testing (possibly dllexport works different?) I don't have objdump available though (didn't want to register for Visual Studio)
Sebastinas
JonnyJD: The libdiscid_EXPORTS stuff can be fixed by including config.h before discid.h in discid.c
That will be part of my commit
Just doing some final testing
JonnyJD
Sebastinas: if you can't get it to work now I recommend getting some sleep and trying it later. I will only test the BSDs and then I probably catch some sleep, too.
