Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 990 991 992 993 994 995 996 997 998 999 1000 .. 5022
— 502,152 events total
2020-11-01 15:25:32 <dminuoso> since it's essentially just advisory locks
2020-11-01 15:25:33 <hpc> just fix those bugs and move on, imo
2020-11-01 15:25:37 <dminuoso> you dont get any real promises about a flock
2020-11-01 15:25:48 <merijn> file locking in linux is so retarded I can't imagine a way windows could be worse, tbh
2020-11-01 15:25:53 <dminuoso> (unless you can somehow prove no other process outside your control will touch your file)
2020-11-01 15:25:56 <dminuoso> merijn: indeed.
2020-11-01 15:26:10 <hpc> merijn: you'll love the discussion on that 110 issue then :D
2020-11-01 15:26:26 → LKoen joins (~LKoen@127.251.88.92.rev.sfr.net)
2020-11-01 15:26:28 <merijn> hpc: Is it *bad* or is it *different*?
2020-11-01 15:26:49 <maerwald> you just retry deletion until your virus scanner is finished
2020-11-01 15:27:12 <hpc> one of the edge cases of one way of deleting a directory holds onto that file path until every handle to it is closed
2020-11-01 15:27:31 <hpc> so you can't make a new file with that name, even though the name isn't currently present on the filesystem
2020-11-01 15:27:37 <dminuoso> maerwald: Anyhow. If your argument is `unix lacks higher level API` that doesnt make unix bad, perhaps the desire to have "high level APIs" or worse "high level cross-platform APIs" is doomed from the get go
2020-11-01 15:27:39 <Uniaika> 1/
2020-11-01 15:27:47 <Uniaika> (woops)
2020-11-01 15:27:53 <maerwald> dminuoso: yes, I wrote that high-level API
2020-11-01 15:27:59 <dminuoso> Which package?
2020-11-01 15:28:18 <Uniaika> oh btw, `transpose` is now leak-free thans to the work of David Feuer and myself
2020-11-01 15:28:37 <dminuoso> merijn: Did you order a DNA sample of dfeuer by the way?
2020-11-01 15:28:38 <maerwald> dminuoso: hpath-directory on RawFilePath and hpath-io on Path
2020-11-01 15:28:41 <dminuoso> We need him cloned.
2020-11-01 15:29:02 <merijn> hpc: You can't say something useful about a single operation like that without considering the design that lead to it. Like "is files being open blocking deletion" a bad design? Not necessarily, imo. It's different than what people are used to on unix, but there's definitely sense to it
2020-11-01 15:29:20 <hpc> that's different
2020-11-01 15:29:30 <hpc> this is files being deleted blocking creation
2020-11-01 15:29:35 <dminuoso> At the end filesystems are complex beasts that I really dont like working with.
2020-11-01 15:29:37 <Uniaika> dminuoso: for the moment I'm acting as a surrogate, I'm the one who put together his patch and test :P
2020-11-01 15:29:41 <dminuoso> Relational databases are so much more pleasant.
2020-11-01 15:29:49 <hpc> the file is already gone from the filesystem
2020-11-01 15:29:53 <merijn> hpc: Is it? AFAICT the deletion is blocked because someone has the file open and locked
2020-11-01 15:29:56 <hpc> it's not present and locked
2020-11-01 15:30:02 → invaser joins (~Thunderbi@31.148.23.125)
2020-11-01 15:30:32 <dminuoso> Clearly we need filesystems with STM semantics and get rid of these silly race conditions.
2020-11-01 15:30:34 <merijn> At what point is a file "no longer present" in your opinion?
2020-11-01 15:31:10 <merijn> dminuoso: This is why I tell people they should just use SQLite for all their applications, then you can just let their code deal with it and that's better tested than yours anyway :p
2020-11-01 15:31:32 <geekosaur> so a file's path is part of its metadata on windows but not on unix. this is a reasonable choice imo
2020-11-01 15:31:38 × cosimone quits (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd) (Quit: cosimone)
2020-11-01 15:31:41 <maerwald> merijn: https://www.sqlite.org/src/info/89f1848d7f
2020-11-01 15:32:02 <merijn> maerwald: That does not in anyway affect the truth of my statement
2020-11-01 15:32:05 <dminuoso> merijn: Indeed, Im a big fan of sqlite myself too.
2020-11-01 15:32:09 <merijn> I didn't say it was flawless
2020-11-01 15:32:19 <merijn> I said it was more battle tested than anything any of us have ever written
2020-11-01 15:32:21 <maerwald> merijn: no, this was just an example of windows
2020-11-01 15:32:30 <maerwald> and that sqlite retries
2020-11-01 15:32:41 <merijn> maerwald: I am still not convinced that's necessarily *bad* design
2020-11-01 15:32:43 <maerwald> see `winDelete`
2020-11-01 15:32:47 <maerwald> I didn't say it is
2020-11-01 15:33:08 <merijn> Right, but hpc (and lots of people when windows comes up) do
2020-11-01 15:33:30 <dminuoso> Anyhow. Maybe someone here knows, Im looking for a package to facilitate generating multiple Haskell modules, but ideally with utilities like `reify` that could let me reflect on already generated code.
2020-11-01 15:33:33 <merijn> afaict most of the "windows is bad" people just mean "windows != posix and I hate having to read docs"
2020-11-01 15:33:38 × fl0_id quits (~fl0_id@claudia.s7t.de) (Quit: ZNC)
2020-11-01 15:33:38 × MTwister quits (~Twister@claudia.s7t.de) (Quit: Servus!)
2020-11-01 15:34:46 <merijn> Windows is (at worst) of "average engineering quality comparable to linux, etc.. More realistically, imo, the engineering of most of windows is orders of magnitude better than linux
2020-11-01 15:35:31 <maerwald> here's anothe example of haskell code handling windows file deletion issues https://github.com/input-output-hk/cardano-wallet/blob/master/lib/core/src/Cardano/DB/Sqlite/Delete.hs
2020-11-01 15:35:31 × chkno quits (~chkno@75-7-2-127.lightspeed.sntcca.sbcglobal.net) (Read error: Connection reset by peer)
2020-11-01 15:35:44 <dminuoso> merijn: I still think much of the bad reputation windows has had, is millions of uneducated users installing shitty drivers, and then blaming microsoft for BSODs.
2020-11-01 15:36:05 <merijn> dminuoso: Yes
2020-11-01 15:36:07 → chkno joins (~chkno@75-7-2-127.lightspeed.sntcca.sbcglobal.net)
2020-11-01 15:36:17 <merijn> dminuoso: Also, just all drivers being shitty
2020-11-01 15:36:30 <maerwald> I get xserver crashed with nvidia drivers
2020-11-01 15:36:31 → troydm joins (~troydm@unaffiliated/troydm)
2020-11-01 15:36:47 <merijn> IMO, the biggest mistake in Windows engineering for 95/98 was MS assuming that every 3rd party vendor would be as careful with code running in kernel space as they were
2020-11-01 15:37:02 <maerwald> but then again, xserver is one of the worst codebases out there :p so not sure I'd blame nvidia
2020-11-01 15:37:07 → Saukk joins (~Saukk@2001:998:f9:2914:1c59:9bb5:b94c:4)
2020-11-01 15:37:22 <dminuoso> Speaking of x, its sad to see its developed has essentially completely stopped
2020-11-01 15:37:38 <dminuoso> Since RH has switched to wayland, that's the last industry funding/work gone.
2020-11-01 15:37:39 <dminuoso> :(
2020-11-01 15:38:48 <maerwald> wayland has no concept of window icons :p
2020-11-01 15:39:03 <maerwald> window managers have to implement a lot of logic to get window icons
2020-11-01 15:39:17 <dolio> What does this have to do with Haskell?
2020-11-01 15:39:32 <dminuoso> dolio: `directory` is bad.
2020-11-01 15:39:34 <dminuoso> :p
2020-11-01 15:39:41 <dolio> Wrong.
2020-11-01 15:39:51 <dminuoso> dolio: No, that's how this all started.
2020-11-01 15:39:56 <dolio> I mean, that may be true, but it has nothing to do with Wayland.
2020-11-01 15:41:15 <hekkaidekapus> We might get a wayland-monad as a successor to xmonad from the discussion :P
2020-11-01 15:41:42 <dolio> I thought someone mentioned that already existed somewhere a while back.
2020-11-01 15:41:51 <dminuoso> hekkaidekapus: waymonad already is a thing
2020-11-01 15:41:55 <hekkaidekapus> Ah, dunno!
2020-11-01 15:41:57 <dminuoso> ttps://github.com/waymonad/waymonad
2020-11-01 15:42:34 <hekkaidekapus> hehehehe… The discussion is doomed then :D
2020-11-01 15:42:35 <dminuoso> Though work on that one has dried up too
2020-11-01 15:43:05 → ManiacTwister joins (~Twister@2a01:4f8:171:4de::40:2)
2020-11-01 15:43:25 × xerox_ quits (~xerox@unaffiliated/xerox) (Ping timeout: 240 seconds)
2020-11-01 15:43:28 → heatsink joins (~heatsink@107-136-5-69.lightspeed.sntcca.sbcglobal.net)
2020-11-01 15:43:42 <maerwald> also, my impression of `directory` was that it wasn't written with POSIX or freedesktop spec in mind, so things like https://github.com/haskell/directory/issues/102 pop up now and then
2020-11-01 15:44:09 → fl0_id joins (~fl0_id@2a01:4f8:171:4de::40:2)
2020-11-01 15:44:53 → dyeplexer joins (~lol@unaffiliated/terpin)
2020-11-01 15:45:19 <dminuoso> Im unsure what your point really is. The bug was fixed.
2020-11-01 15:45:26 <maerwald> yes, I fixed it :p
2020-11-01 15:45:31 hackage ghc-lib-parser 0.20201101 - The GHC API, decoupled from GHC versions https://hackage.haskell.org/package/ghc-lib-parser-0.20201101 (shayne_fletcher)
2020-11-01 15:46:11 <maerwald> but what about other functions
2020-11-01 15:46:30 hackage ghc-lib 0.20201101 - The GHC API, decoupled from GHC versions https://hackage.haskell.org/package/ghc-lib-0.20201101 (shayne_fletcher)
2020-11-01 15:46:34 <maerwald> if your main concern is cross platform, I think these things will happen
2020-11-01 15:46:43 × ech quits (~user@gateway/tor-sasl/ech) (Ping timeout: 240 seconds)
2020-11-01 15:48:59 <maerwald> your mental model is the intersection of windows and linux, not POSIX
2020-11-01 15:49:23 → lagothrix joins (~lagothrix@unaffiliated/lagothrix)
2020-11-01 15:53:06 × Franciman quits (~francesco@host-79-36-167-172.retail.telecomitalia.it) (Quit: Leaving)
2020-11-01 15:56:20 × dysfigured quits (dysfigured@2600:3c00::f03c:92ff:feb4:be75) (Quit: I hate quit messages.)
2020-11-01 15:56:49 monsterchrom is now known as monochrom
2020-11-01 15:59:13 × bitmapper quits (uid464869@gateway/web/irccloud.com/x-stehtktsnkxbdlii) (Quit: Connection closed for inactivity)
←Prev  Next→
Page 1 .. 990 991 992 993 994 995 996 997 998 999 1000 .. 5022

All times are in UTC.