ActsAsTree -- Extends ActiveRecord to add simple support for organizing items into parent–children relationships.
HTML5/Websockets/Eventmachine Bomberman
Define enum classes in Rails and use them to enumerate ActiveRecord attributes
The administration framework for Ruby on Rails applications.
A pure Unix shell script implementing ACME client protocol
felixbuenemann/active_model_serializers 0
ActiveModel::Serializer implementation and Rails hooks
felixbuenemann/acts_as_ordered_tree 0
ActiveRecord extension for ordered adjacency trees support
felixbuenemann/acts_as_solr_reloaded 0
ActsAsSolr with new features
ActsAsTree plugin
pull request commentxbmc/xbmc
Fix Linux GL Renderer color primaries detection
Btw. why was the "Backport: Needed" label removed?
Since #17109 was backported to Leia, this should backport fine.
comment created time in an hour
push eventfelixbuenemann/xbmc
commit sha cf763cadf4fafbfd639097e83511f00132bcaeee
Merge pull request #17680 from AlwinEsch/cleanup-part [addons][audioengine] cleanup the addon audioengine to match "C", match clang and to make more safe.
commit sha c9f6d45c9d4f4b639c1759b188969645189cb9d7
[video][PVR] Recordings: Add support for 'Play from here' context menu item.
commit sha 35ce7e7efe7ecf74f98cf539e0196301d9e26fd3
[video] Fix playlist window not playing the clicked item if 'autoplay next item' is enabled for 'Uncategorised' content in Kodi settings.
commit sha 9e72512102a9644a3428d418c5efba291d53a973
[Estuary][PVR] Recordings window: Add 'Go to playlist' to side blade.
commit sha 0d42fdc5ac22a624596ca5736bfd1d71a10cdf1b
[Estuary] Playlist window: Add support for PVR recordings.
commit sha 793a7af00326ad000ee62784c07ff7d42c7a73ff
Merge pull request #17703 from enen92/sortaddonupdates [Addons] Ensure addon installation order when migrating to matrix
commit sha a1f431262b7d7964cd500a2619938906c61643ec
Merge pull request #17809 from ksooo/pvr-recordings-playlist-actions [PVR][Estuary] Recordings Playlist Actions
commit sha eb7ab2d6d680c4c7047beb062b375086b5b9637c
Dont send referer on redirects. Same behaviour as ffmpeg ans browsers
commit sha 16efbd549adc9db452460eb81836cb19bc42f9d2
Fix for <fanartres> not being used in advancedsettings.xml
commit sha 90d5269b5a0ee455cacd29fae9690c61bf013d5e
Merge pull request #17816 from matthuisman/ffmpeg_select_program_fix [videoplayer] Fix HLS selecting lowest quality stream & Live HLS always play from start
commit sha e079b8f54058e043e2e8a71920a0c915f46225d4
Merge pull request #17801 from vpeter4/uhd_br [bluray] Correctly recognize subtitle language
commit sha 780ed08ef6015e619e65ba5d2bb31e02e85588a4
[video] ** CID 1463261: Resource leaks (RESOURCE_LEAK) /xbmc/video/ContextMenus.cpp: 171 in CONTEXTMENU::<unnamed>::AddRecordingsToPlayListAndSort(const std::shared_ptr<CFileItem> &, CFileItemList &)()
commit sha 5b4868e1e15c3c3d714f7dd91ff2bf09425c381e
Merge pull request #17734 from bkuhls/master-cmake-memfd_create cmake: fix check for memfd_create
commit sha e378fa07076af1966e6c60916a1b66f27557ebaa
Merge pull request #17652 from djp952/issue17623-httpdirectory [Filesystem] HTTP Directory parser enhancements for multi-line anchor tags
commit sha 20a703422aa99c02a2a5d6116fe5b5ef566bf961
Merge pull request #17837 from ksooo/coverity-fix [video] ** CID 1463261: Resource leaks (RESOURCE_LEAK)
commit sha b92471a6b2c784b2ef599251b1cbb399b515eb51
Only check if items are .iso and expand disc image when in file view
commit sha 0ff666612fa11c840cb6a75758ab79cbd3d634be
[skins] fix sortletter
commit sha 875a8f65805b82fa95e902ec194f8a4673a9b3e1
Merge pull request #17843 from DaveTBlake/ISOCheckNotLib [Fix]Only expand iso images in music file view window to avoid GUI freezes
commit sha 12e24d3cbeb120684c19766b0469bc21247bdfa0
Use poster before thumbnail in Top Shelf
commit sha c4bb6ab9e9b44a348c3708d31ed30ccd125c1d54
[PVR] Optimize EPG Grid Container Model.
push time in 2 hours
pull request commentxbmc/xbmc
Fix Linux GL Renderer color primaries detection
@fritsch Rebased on master at b7e053aebc and fixed clang-format warnings.
comment created time in 2 hours
pull request commentDragonCherry/AssetsPickerViewController
Added small update to README.md to mark SPM support as done.
comment created time in 4 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha c23fd76727669d12bff798c678171816b8c2e338
Add SwiftPM manifest and localized string lookup
push time in 4 hours
PR opened DragonCherry/AssetsPickerViewController
This makes AssetsPickerViewController compatible with SwiftPM.
I've dropped the Device depedency, since there were only about 10 LOC used and it seems unmaintained with multiple open PRs to add SPM support among other things.
pr created time in 4 hours
pull request commentxbmc/xbmc
Fix Linux GL Renderer color primaries detection
Sure, I can rebase this later.
Can't say I'm a fan of those automated style checks, most of them decrease readability instead of improving it and they also make it harder to see what actually changed in the diff…
comment created time in 5 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha f34bf7bdf6fa91fd3a749b8e3b04a6665cc4fe61
Remove dependency on Device library
commit sha 5bb91fd452a81c7e4fb89ccf15ddb094b469f666
Add SwiftPM manifest and localized string lookup
push time in 8 hours
created tagfelixbuenemann/AssetsPickerViewController
Powerfully Customizable - Multiple Photo & Video Picker Controller
created time in 9 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha b666bd33bb919af76a2079e728a429bca8477f8e
Remove dependency on Device library
commit sha 7c4009b85f828b4cb03424dd72a4d8ce3980d7be
Remove SwiftPM dependency on Device
commit sha 1211bc275c92dd4984048b7571cb0e38a18d896d
Remove Device lib from podspec and example
commit sha 906115ab1cbc62ff33d032c7a5926173bfcccfa7
Merge branch 'remove-device-framework'
commit sha 2c315f4586880b591d2e5065326289e8da839932
Bump version to 3.0.0
push time in 9 hours
create barnchfelixbuenemann/AssetsPickerViewController
branch : remove-device-framework
created branch time in 9 hours
created tagfelixbuenemann/AssetsPickerViewController
Powerfully Customizable - Multiple Photo & Video Picker Controller
created time in 11 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha 4b7524412e0212b6c64a2b4e879930b651ce186a
Bump version to 2.9.7
push time in 11 hours
push eventfelixbuenemann/BarcodeScanner
commit sha 5c57ffc31fdd3c0d0328f71e62176c9c33599b61
Add Polish localisation
commit sha afc717480f9e3c9fceb15f00521f9d02346d742e
Merge pull request #74 from stolkachov/master Add Polish localisation
commit sha c2dbf967dfa680475c0443675886a0e570b53b1a
Improved iPhoneX support, fixed simulator crash
commit sha 3f2cf4f893e465a33bf84d8ec5c6289e1fe376ed
Fixed unused variable warning
commit sha ecfe4a09ddfcbed860f24fdd4ec1b14a850772b9
Fixed indentation, UIScreen
commit sha 19f6c2c8cc0e247c9f63a67e1c7f76c5c9476aa9
Merge pull request #77 from prcodes/develop Fixing iPhone X issues, Simulator crash
commit sha 0b665d84ce284bd8dfa4a616972781a5d2d338c8
Refactor access control
commit sha 922a0e59ee9cbd9a5b931647ee2895cd2618cf13
Bump version number
commit sha 3819e678d51dae9615b86ce6e6d56c32269e4326
Update podspec
commit sha 7564207a7492b365d81e5b6e1949978c6646d3e6
Merge pull request #78 from hyperoslo/refactor/access-control Refactor: access control
commit sha 570806f90f4ee1b14aca6f15e8ebbfd4d7e8ce34
Purely return views
commit sha 57c68b18d54fd51455d4702bb669eb8bea522c44
Use UINavigationBar
commit sha 508aba959343701b3169236595f87e0f976ef809
Merge pull request #82 from hyperoslo/fix/bar Use UINavigationBar
commit sha b3e52ba3474ddd860fe7beac3ce62c87fe4ff778
Bump to 3.0.2
commit sha 05af1b0918a947910dfddf8c6d2dc08d9ec59862
Use topLayoutGuide
commit sha 7e13c70514e875a2c9c354b78042d3ce48576d07
Add sizeToFit
commit sha 4581038f648f349f3e0efd9ced9a21bf46ecb7e8
Merge pull request #84 from hyperoslo/fix/bar Fix title and button on navigation bar in iOS 10
commit sha 3617d6ca615700fbf9c2d9df71a5664170e96e96
Bump to 3.0.3
commit sha 7f51fe539ab317a23f17cd06a47d18b61dc46a85
Refactor header as a child view controller
commit sha c172beeb6a45f6c2bbc9931bcb5da2c7f741e03e
Refactor header view controller
push time in 11 hours
push eventfelixbuenemann/TinyLog
commit sha bf82e7541fc560e45e96a7e7f62cd96a7818602a
Fix Xcode 9.1 deprecation warning for characters This fixes the following warning in Xcode 9.1 and later: 'characters' is deprecated: Please use String or Substring directly
commit sha 3a8553d0161921e40a028017a578ff27e779d790
Fix off-by-one error for function name truncation There was an error when this code was converted from Swift 3 to 4, because `substring(to: index)` is exclusive, but `...index` is inclusive, so we need to use the exclusive `..<index` syntax.
commit sha e34ccc927a062f27bc234d0d0570128cb6c90b0f
Merge pull request #2 from felixbuenemann/fix-xcode-9.1-characters-deprecation-warning Fix Xcode 9.1 characters deprecation warning and off-by-one error
commit sha 4192df38ff5238196886ab6cae251c9f39a70539
bumped version to 2.0.5
commit sha 69221188ab32e74ec966c6359e0fdcd15e3dc974
added flag variables for enabling/disabling logs
commit sha ab4cc3e7d499c31e150272603c8e6b58f471b4a9
Merge branch 'release/2.0.6'
commit sha 5aac466ac2fd042f85a054a8cd817934e51efd1d
patch for flag type
commit sha cf76d3d0cb6046f9602c14594d5256215582c38b
Merge branch 'release/2.0.7'
commit sha a2ede89be068d88b8d1f3309fdd72875b7e0fe7a
Fix Swift 5 deprecation warnings
commit sha 7103e3bb21472efef2754f5934c8e0fd60154d04
Add SwiftPM manifest
commit sha 05b60a7683314984ed20a8aa326b3b30210fe3a5
Bump version to 2.0.8
push time in 11 hours
push eventfelixbuenemann/SimpleExif
commit sha 969a4d8de3e43d2f35dc8fd5edb632c225119c86
Bump version to 0.0.4
push time in 11 hours
push eventfelixbuenemann/SimpleExif
commit sha b991eba3f2dcb48f058667e2971bcf416102804f
Fix podspec license
commit sha 14be973dd2e5a99f02d85fadab1fa631f751ce33
fix exif key value
commit sha 923e039908378ff06f4ddbcd7b039acde74bd78d
Fix macOS build warnings
commit sha 65f891bf514532f80dbdd8cc7cb2363a69993a5b
Add SwiftPM manifest and OS build guards
push time in 11 hours
create barnchfelixbuenemann/SimpleExif
branch : swift-package-manager
created branch time in 11 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha a3d887e50915907288068cfbfe7065a1b0e4b9bc
Add SwiftPM manifest and localized string lookup
push time in 11 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha a3d887e50915907288068cfbfe7065a1b0e4b9bc
Add SwiftPM manifest and localized string lookup
push time in 11 hours
push eventfelixbuenemann/Device
commit sha 37ea80a827c4b8570c78e5ca34561d6a70c61234
Update device support tables
commit sha 6a61b1bdee21dc94e30df0af428ab40ffbd284fe
Merge branch 'update-ios-device-tables'
push time in 11 hours
push eventfelixbuenemann/Device
commit sha 37ea80a827c4b8570c78e5ca34561d6a70c61234
Update device support tables
push time in 11 hours
PR opened Ekhoo/Device
This adds support for the following devices:
- iPhone (2G)
- iPhone 3G
- iPhone 3GS
- iPhone SE 2
- iPad Mini 5
- iPad 8
- iPad Air 4
- iPad Pro 11" 2
- iPad Pro 12,9" 4
- iPod Touch 7
The following screen sizes are added:
- iOS 10.9 Inch (iPad Air 4)
- macOS 16 Inch (MacBook Pro)
pr created time in 11 hours
created tagfelixbuenemann/Device
Light weight tool for detecting the current device and screen size written in swift.
created time in 12 hours
push eventfelixbuenemann/Device
commit sha e48952b9e3fe51e84d0ec1c9fa70daaceeaf5d81
Update device support tables
commit sha dd5504dfbaf609274bb0b28a552385bdc967a724
Merge branch 'update-ios-device-tables'
push time in 12 hours
create barnchfelixbuenemann/Device
branch : update-ios-device-tables
created branch time in 12 hours
push eventfelixbuenemann/Device
commit sha 74b5306d524dda8b03ab1a12df7d79d05620d721
Add SwiftPM manifest and OS source guards
push time in 14 hours
pull request commentEkhoo/Device
Add SwiftPM manifest and OS source guards
Ooops, just noticed there are a bunch of other PRs that do the same, like #93 and #100.
I noticed that #93 uses canImport which should be better thatn the os(iOS) check.
comment created time in 14 hours
PR opened Ekhoo/Device
This allows usage of Device with Swift Package Manager.
Some build conditions are added to allow a single product and target for iOS and macOS.
pr created time in 14 hours
Light weight tool for detecting the current device and screen size written in swift.
fork in 14 hours
push eventfelixbuenemann/AssetsPickerViewController
commit sha 80532057c3d0d6c07de94e64cb3a359b69e9b463
Add missing UIKit imports
commit sha c981881f1fad4b4ca5522ac3754ac58792748899
Add SwiftPM manifest and localized string lookup
push time in 14 hours
create barnchfelixbuenemann/AssetsPickerViewController
branch : swift-package-manager
created branch time in 14 hours
fork felixbuenemann/AssetsPickerViewController
Powerfully Customizable - Multiple Photo & Video Picker Controller
fork in 14 hours
issue commentteamhephy/workflow
Simple installation of Hephy wouldn't work. Client on Mac OS X and Cluster on AWS
@dmcnaught No I haven't, I'm still on 1.17.
comment created time in a day
issue commentamerine/acts_as_tree
Can't Process Ordering on a column named as order
Have you tried order: :order instead?
comment created time in 3 days
issue closedaskn/progress
Memory leak with high number of ticks
I tried this shard in a project that worked through about 11 million rows.
I set bar.total to the number of rows and at about 50% memory usage had grown to about 30 GB.
After removing the progress bar in replacing it with a print statement the memory usage dropped 15 MB.
crystal --version
Crystal 0.35.1 (2020-06-19)
LLVM: 10.0.0
Default target: x86_64-apple-macosx
Similar code ("do stuff" sends line to a channel):
require "progress"
File.open("data.txt", "r") do |f|
lines = 0
f.each_line { lines += 1 }
f.rewind
bar = ProgressBar.new
bar.total = lines
f.each_line.with_index do |line, index|
# do stuff
bar.inc
end
bar.done
end
closed time in 5 days
felixbuenemann
issue commentaskn/progress
Memory leak with high number of ticks
I think the problem wasn't caused by the progress bar, since another run seem to also have spiked in memory usage after 50% through the dataset, so I'm closing this.
comment created time in 5 days
issue openedaskn/progress
Memory leak with high number of ticks
I tried this shard in a project that worked through about 11 million rows.
I set bar.total to the number of rows and at about 50% memory usage had grown to about 30 GB.
After removing the progress bar in replacing it with a print statement the memory usage dropped 15 MB.
crystal --version
Crystal 0.35.1 (2020-06-19)
LLVM: 10.0.0
Default target: x86_64-apple-macosx
Similar code ("do stuff" sends line to a channel):
require "progress"
File.open("data.txt", "r") do |f|
lines = 0
f.each_line { lines += 1 }
f.rewind
bar = ProgressBar.new
bar.total = lines
f.each_line.with_index do |line, index|
# do stuff
bar.inc
end
bar.done
end
created time in 7 days
push eventfelixbuenemann/SimpleExif
commit sha c0e7ec83334922e40cbf94074d9ba6708b82ad5b
Fix license filename in podspec
push time in 14 days
startedlian-yue/vue-upload-component
started time in 15 days
push eventfelixbuenemann/plexdrive
commit sha 3175c9feaf1637c9bbd4befd99744d1e125e719d
Remove redundant len in stack
push time in 20 days
push eventfelixbuenemann/plexdrive
commit sha 59e461ff069e05878ed7c9d7af0573cea3d32abc
Fix help description for max-chunks option (#372) Fixed description for --max-chunks stating "maximum number of chunks to be stored on disk" when it actually affects "maximum number of chunks to be stored in memory".
commit sha cb8424cfd89dbdffa483a6f9e81e0c4f2173cb9a
Update build badge from Travis CI to GitHub Actions (#379)
commit sha c24094dd3f37745f0736e4588d1ebe853c097688
Manual cache refresh (#380) * Implement SIGHUP handler, fixes #176 * Mention signals in README Co-authored-by: Auri <[email protected]>
commit sha 58dc59fb82c76d6a1f81050ca6099eb457264f3f
Skip store of existing buffer
commit sha b75550e39dc032995c189b4ed16de635691b66a2
Allow parrallel reads from storage using rw lock
commit sha ff346d9a6ce34f8e3bcdbab8d590cdc45f3b54ef
Use normal simple write lock for stack since there are no read only locks.
commit sha 9a3a66ca13ceacd38e62d9ca189649aee4bfbeef
Mark manager queue entry as private
commit sha 836060872abd8f7ccf9a49305eaf796c57e34bd1
Add buffer pool for download bodies and chunks This should reduce memory allocation and gc pressure.
commit sha 5c01baa29d98e468a911b199cef0498c16e7175e
Preallocate buffer pool buffers This should avoid many reallocation on the buffers.
commit sha 5ac70e917633ce1c1db6545526fbe50447cb9393
Log buffer pool usage and size
commit sha 86c2ae9e7c09904e7286fcc6d05035876362237f
Remove bytes.Buffer from buffer implementation Instead just implement the minimal interface we need.
commit sha 50f69918829999f9f0f4ea9c967cd482fb9e86ed
Limit buffer pool size Since the sync.Pool tends to grow way beyond max buffers, we wait until a slot becomes free.
commit sha 6baa3fe998e6754a77619d3b11861fbbbc6b6264
Replace sync.Pool with pre-allocated chan pool
commit sha e8120923866ba8a02c6ecc06532d83ddcc1fe2ab
Try to pre-allocate buffer memory
commit sha e39cab176fb2add02a859edee068c00b70f9dce6
Only store each chunk once from downloader
commit sha d7af8334bfd83783310942c8eb89931d7fcb15c0
Simplify buffer pool, remove ref-counting
commit sha 4eb58e3c747fce6e153a64be320295e630cc2b80
Rename buffer pool source
commit sha 29e26fa341cc25f4c97fe0d31c2b55dba22bb450
Crash if buffer pool deadlocks This should never happen if the pool size is correctly calculated, so if it does happen crash to help with debugging and allow to recover from deadlock.
commit sha ce4ac34da142fec5488ed23220445cadd08b61f8
Fix buffer leak when chunk is already stored
commit sha d12bdf84f067bae447c70d7c944b029239d59587
Fix buffer leak when download fails
push time in 20 days
push eventplexdrive/plexdrive
commit sha c24094dd3f37745f0736e4588d1ebe853c097688
Manual cache refresh (#380) * Implement SIGHUP handler, fixes #176 * Mention signals in README Co-authored-by: Auri <[email protected]>
push time in 20 days
PR merged plexdrive/plexdrive
This implements a manual cache refresh py sending a SIGHUP to the plexdrive process.
Credit for the code goes to @aurieh, I've only amended the README.
Fixes #176
pr closed time in 20 days
issue closedplexdrive/plexdrive
[ Feature Request ] PlexDrive update cache on SIGHUP
Could you please add ability to manually trigger cache update with SIGHUP For example: kill -SIGHUP $(pgrep -f "plexdrive") would update cache.
Something similar like rclone added here: https://github.com/ncw/rclone/commit/58a82cd5780a91cf3adfb043d476a752688d4f32
With that we could force refresh with custom scripts when we know changes happened and not use some small refresh timers.
closed time in 20 days
ajkis
PR opened plexdrive/plexdrive
This implements a manual cache refresh py sending a SIGHUP to the plexdrive process.
Credit for the code goes to @aurieh, I've only amended the README.
Fixes #176
pr created time in 20 days
issue commentplexdrive/plexdrive
Do you get any of these errors, after removing the cache.bolt and restarting plexdrive with -v 3?
- "Could not get changes"
- "Ignoring change type"
- "Could not map Google Drive file"
Everything that looks suspicious in the logs between the lines "First cache build process started..." and "First cache build process finished!" will be usefull for debugging.
comment created time in 20 days
issue commentplexdrive/plexdrive
This has been discussed before, it is up to the fuse package used to implement Windows support.
Currently bazil.org/fuse doesn't support Windows, if you know a Go FUSE package that does, let me know.
comment created time in 20 days
issue commentplexdrive/plexdrive
Do you get anything interesting in the logs, if you rename the cache.bolt file and restart plexdrive with the -v 3 flag?
Does ist always show the same incomplete list of files or does it change when you rebuild the cache by renaming or removing cache.bolt?
comment created time in 20 days
pull request commentplexdrive/plexdrive
Add direct_io mount option, max_readahead default
You can enable it with -o direct_io.
There are multiple reasons it is not enabled by default:
- It emulates the libfuse mount option by the same name and you couldn't disable it, if it was on by default.
- Direct-IO is primarily a Linux thing, I'm not sure how it behaves in other platforms like FreeBSD
- With the new default of 128 KB for
max_readahead, overhead for buffered reads is greatly reduced and triggers 128 KBreads instead of the 64 KB reads the Plex Media Server would usually use. - On macOS osxfuse uses up to 16 MB for buffered reads.
That said I have been running with direct_io enabled on Linux for many months without problems.
To do a real side by side benchmark against 128 KB buffered IO, I would need a very high bitrate stream (around 100 MBits) with an almost completely constant bitrate, because bitrate fluctuations cause huge differences in CPU load, making comparisons hard.
I am currently using a 4K stream with up to 180 MBits peak bitrate for testing on a single core Xeon Gold VM with 2GB RAM, but is has pretty heavy bitrate fluctuation.
comment created time in 20 days
push eventfelixbuenemann/plexdrive
commit sha 3622aa47caf390a36301b5a05d09f9826fdf4db1
Fix mmap disk cache always enabled
commit sha ded386705bf1813db78e8e85c882a3b6836cf392
Enable zero-copy support (requires patched fuse pkg)
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 3622aa47caf390a36301b5a05d09f9826fdf4db1
Fix mmap disk cache always enabled
push time in 21 days
push eventplexdrive/plexdrive
commit sha cb8424cfd89dbdffa483a6f9e81e0c4f2173cb9a
Update build badge from Travis CI to GitHub Actions (#379)
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 6b5ad99245c6a12d905e2fdfcc399544eae21d21
Enable zero-copy support (requires patched fuse pkg)
push time in 21 days
pull request commentplexdrive/plexdrive
Fix help description for max-chunks option
Thanks!
comment created time in 21 days
push eventplexdrive/plexdrive
commit sha 59e461ff069e05878ed7c9d7af0573cea3d32abc
Fix help description for max-chunks option (#372) Fixed description for --max-chunks stating "maximum number of chunks to be stored on disk" when it actually affects "maximum number of chunks to be stored in memory".
push time in 21 days
PR merged plexdrive/plexdrive
Fixed description for --max-chunks stating "maximum number of chunks to be stored on disk" when it actually affects "maximum number of chunks to be stored in memory".
According to Issue#246 --max-chunks represents chunks stored in memory not on disc.
This is backed up by my VM running out of memory when I set --max-chunks to be capable of storing 256GB worth of chunks.
Fixed the typo so that others wont make the same config mistakes and have machines crash :)
pr closed time in 21 days
create barnchfelixbuenemann/plexdrive
branch : mmap-storage-offheap-zero-copy
created branch time in 21 days
delete branch felixbuenemann/plexdrive
delete branch : use-config-path-for-cache-file
delete time in 21 days
push eventplexdrive/plexdrive
commit sha 03a3639155c590981bdc70156bb41dc3b1bd74fa
Put cache.bolt in config path by default (#359) Many users change the config path to run multiple copies of plexdrive, but forget to also set the cache file path, which results in corruption of the cache.bolt database. This changes the default for the cache file, so it is located in the config path unless otherwise specified.
push time in 21 days
PR merged plexdrive/plexdrive
Many users change the config path to run multiple copies of plexdrive, but forget to also set the cache file path, which results in corruption of the cache.bolt database. This changes the default for the cache file, so it is located in the config path unless otherwise specified.
pr closed time in 21 days
push eventfelixbuenemann/plexdrive
commit sha b28276bea482289b1c43e5e35e5f44d4804bc8f6
Put cache.bolt in config path by default Many users change the config path to run multiple copies of plexdrive, but forget to also set the cache file path, which results in corruption of the cache.bolt database. This changes the default for the cache file, so it is located in the config path unless otherwise specified.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha b6c8f1709a22cb889fb5975a754abd0bcf93d5ff
Add GitHub Actions Go Workflow (#378) This builds plexdrive and runs the tests.
commit sha 6baff616561d1e2627b7084e346d139731fd7910
Update all github repo references (#377)
commit sha 743ad0f5d5a6f583eefb1e72020e6823fae9b83f
Fix wrong path for GitHub Action Go Workflow
commit sha a687dce6e7af9bb8713176d32f44c1418ec9ca35
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
commit sha 0330ec3ac0529e77c71fb5df93002f4bd8119e1d
Fix reads across chunk boundaries (#350) * Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data. * Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
commit sha 8658cf74dfc8cc9b9da1b5b194db6032321aa4bc
Add direct_io mount option, max_readahead default (#351) This reduces the number of FUSE read requests significantly by using a default max_readahead value of 128 KB, which is the default readahead used by libfuse and also the maximum request size for buffered I/O used by the FUSE implementation in the Linux kernel. We need to set a default, because the bazil.org/fuse package otherwise defaults to max_readahead=0, which disables readahead completely causing FUSE to request only 4 KB (one page) per request. The direct_io option allows to force all files to be opened with the O_DIRECT flag, bypassing kernel cache, I/O reordering and alignment. Depending on the apps I/O patterns this can be benefitial, because it avoids double-buffering of data in both the storage chunk cache and kernel memory leading to reduced memory usage and CPU usage for moving data in and out of kernel caches. It will also theoretically allow FUSE requests of up to 1 MB instead of the maximum of 128 KB, which FUSE enforces for buffered I/O, because it splits larger requests into multiple 128 KB chunks that it requests asynchronously. Currently the maximum with Direct (unbuffered) I/O will still be 128 KB, because the bazil.org/fuse package does not yet support the required FUSE version and max_pages capability required for larger I/Os.
commit sha 114bd3b8f89815d5408b29391ddf0bdb00dcdd5c
Put cache.bolt in config path by default Many users change the config path to run multiple copies of plexdrive, but forget to also set the cache file path, which results in corruption of the cache.bolt database. This changes the default for the cache file, so it is located in the config path unless otherwise specified.
push time in 21 days
pull request commentplexdrive/plexdrive
Fix help description for max-chunks option
@Inrixia Can you also fix the description in main.go?
comment created time in 21 days
push eventplexdrive/plexdrive
commit sha 8658cf74dfc8cc9b9da1b5b194db6032321aa4bc
Add direct_io mount option, max_readahead default (#351) This reduces the number of FUSE read requests significantly by using a default max_readahead value of 128 KB, which is the default readahead used by libfuse and also the maximum request size for buffered I/O used by the FUSE implementation in the Linux kernel. We need to set a default, because the bazil.org/fuse package otherwise defaults to max_readahead=0, which disables readahead completely causing FUSE to request only 4 KB (one page) per request. The direct_io option allows to force all files to be opened with the O_DIRECT flag, bypassing kernel cache, I/O reordering and alignment. Depending on the apps I/O patterns this can be benefitial, because it avoids double-buffering of data in both the storage chunk cache and kernel memory leading to reduced memory usage and CPU usage for moving data in and out of kernel caches. It will also theoretically allow FUSE requests of up to 1 MB instead of the maximum of 128 KB, which FUSE enforces for buffered I/O, because it splits larger requests into multiple 128 KB chunks that it requests asynchronously. Currently the maximum with Direct (unbuffered) I/O will still be 128 KB, because the bazil.org/fuse package does not yet support the required FUSE version and max_pages capability required for larger I/Os.
push time in 21 days
PR merged plexdrive/plexdrive
This reduces the number of FUSE read requests significantly by using a default max_readahead value of 128 KB, which is the default readahead used by libfuse and also the maximum request size for buffered I/O used by the FUSE implementation in the Linux kernel.
We need to set a default, because the bazil.org/fuse package otherwise defaults to max_readahead=0, which disables readahead completely causing FUSE to request only 4 KB (one page) per request.
The direct_io option allows to force all files to be opened with the O_DIRECT flag, bypassing kernel cache, I/O reordering and alignment. Depending on the apps I/O patterns this can be benefitial, because it avoids double-buffering of data in both the storage chunk cache and kernel memory leading to reduced memory usage and CPU usage for moving data in and out of kernel caches.
It will also theoretically allow FUSE requests of up to 1 MB instead of the maximum of 128 KB, which FUSE enforces for buffered I/O, because it splits larger requests into multiple 128 KB chunks that it requests asynchronously.
Currently the maximum with Direct (unbuffered) I/O will still be 128 KB, because the bazil.org/fuse package does not yet support the required FUSE version and max_pages capability required for larger I/Os.
IMPORTANT: Must be merged after #350 to avoid short reads with unaligned offsets.
ToDo:
- [x] Document direct_io in README
pr closed time in 21 days
push eventfelixbuenemann/plexdrive
commit sha f7b31d6107ecc769dcc24e1b6a2071ef1dea7607
Add direct_io mount option, max_readahead default This reduces the number of FUSE read requests significantly by using a default max_readahead value of 128 KB, which is the default readahead used by libfuse and also the maximum request size for buffered I/O used by the FUSE implementation in the Linux kernel. We need to set a default, because the bazil.org/fuse package otherwise defaults to max_readahead=0, which disables readahead completely causing FUSE to request only 4 KB (one page) per request. The direct_io option allows to force all files to be opened with the O_DIRECT flag, bypassing kernel cache, I/O reordering and alignment. Depending on the apps I/O patterns this can be benefitial, because it avoids double-buffering of data in both the storage chunk cache and kernel memory leading to reduced memory usage and CPU usage for moving data in and out of kernel caches. It will also theoretically allow FUSE requests of up to 1 MB instead of the maximum of 128 KB, which FUSE enforces for buffered I/O, because it splits larger requests into multiple 128 KB chunks that it requests asynchronously. Currently the maximum with Direct (unbuffered) I/O will still be 128 KB, because the bazil.org/fuse package does not yet support the required FUSE version and max_pages capability required for larger I/Os.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha e205113c294b8cc4823d6a9d4818bfe816883e00
Remove allow_root option from README This option was removed in #340.
commit sha 7f5abf6fe0eb6c11bd1a703594af7280ccba130e
Fix off-by one error in download range request
commit sha 65ff5d41f985df55b3d30bb1bd65183626472c84
Prepare 5.1.0 release
commit sha b6c8f1709a22cb889fb5975a754abd0bcf93d5ff
Add GitHub Actions Go Workflow (#378) This builds plexdrive and runs the tests.
commit sha 6baff616561d1e2627b7084e346d139731fd7910
Update all github repo references (#377)
commit sha 743ad0f5d5a6f583eefb1e72020e6823fae9b83f
Fix wrong path for GitHub Action Go Workflow
commit sha a687dce6e7af9bb8713176d32f44c1418ec9ca35
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
commit sha 0330ec3ac0529e77c71fb5df93002f4bd8119e1d
Fix reads across chunk boundaries (#350) * Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data. * Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
commit sha 757ac1d92e94701f444ca41e2c9453eb99494329
Add direct_io mount option, max_readahead default This reduces the number of FUSE read requests significantly by using a default max_readahead value of 128 KB, which is the default readahead used by libfuse and also the maximum request size for buffered I/O used by the FUSE implementation in the Linux kernel. We need to set a default, because the bazil.org/fuse package otherwise defaults to max_readahead=0, which disables readahead completely causing FUSE to request only 4 KB (one page) per request. The direct_io option allows to force all files to be opened with the O_DIRECT flag, bypassing kernel cache, I/O reordering and alignment. Depending on the apps I/O patterns this can be benefitial, because it avoids double-buffering of data in both the storage chunk cache and kernel memory leading to reduced memory usage and CPU usage for moving data in and out of kernel caches. It will also theoretically allow FUSE requests of up to 1 MB instead of the maximum of 128 KB, which FUSE enforces for buffered I/O, because it splits larger requests into multiple 128 KB chunks that it requests asynchronously. Currently the maximum with Direct (unbuffered) I/O will still be 128 KB, because the bazil.org/fuse package does not yet support the required FUSE version and max_pages capability required for larger I/Os.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 0330ec3ac0529e77c71fb5df93002f4bd8119e1d
Fix reads across chunk boundaries (#350) * Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data. * Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days
push eventplexdrive/plexdrive
commit sha 0330ec3ac0529e77c71fb5df93002f4bd8119e1d
Fix reads across chunk boundaries (#350) * Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data. * Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days
PR merged plexdrive/plexdrive
When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data.
This also removed the float conversions in adjustResponseChunk.
IMPORTANT: Must be merged after #349 to avoid corruption, since we would repeat the off-by-one byte.
ToDos:
- [x] Add unit tests for splitRequestChunks
pr closed time in 21 days
push eventfelixbuenemann/plexdrive
commit sha a687dce6e7af9bb8713176d32f44c1418ec9ca35
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
commit sha 47989bb683cdcca204a4144fc7330b26f06bc529
Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data.
commit sha 8d4122af29b7b033e8f373d0c9d7a2ad2c25139b
Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days
push eventplexdrive/plexdrive
commit sha a687dce6e7af9bb8713176d32f44c1418ec9ca35
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha a687dce6e7af9bb8713176d32f44c1418ec9ca35
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 42767393edf91cb087ca570657f4cc2901d4c265
Add GitHub Actions hack to work without go modules This symlinks the checkout dir to the proper src location in the GOPATH to avoid downloading the master version of subpackages.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 743ad0f5d5a6f583eefb1e72020e6823fae9b83f
Fix wrong path for GitHub Action Go Workflow
commit sha fb8fef174ec79cc9432ce55ddf31310dd397a1ab
Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data.
commit sha 051544d4bc6a0d9336b166933f83720b178cd3c9
Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days
push eventplexdrive/plexdrive
commit sha 743ad0f5d5a6f583eefb1e72020e6823fae9b83f
Fix wrong path for GitHub Action Go Workflow
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 743ad0f5d5a6f583eefb1e72020e6823fae9b83f
Fix wrong path for GitHub Action Go Workflow
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha f7014a82bfbb4a1e6d5b653d549bd0deebd97d3a
Fix option parsing to work with 1 CPU core Several options are based on the number of CPUs, which in some cases lead to option default being set to zero: 1 CPU / 2 => 0 (integer division) 1 CPU - 1 => 0 This fixes the problem by ensuring those default are at least one.
commit sha 6e05ca4a70dec9d2d55a938a50ec2da77582dfc8
Merge pull request #250 from felixbuenemann/fix-plexdrive-1cpu-support Fix option parsing to work with 1 CPU core
commit sha 2acbdd22393035d63a920f5c406ec687a54c3149
Fallback to HOME for getting home directory (#268) Fixes #263
commit sha 4366e54d3284980b670a6bd2e1463df338f67de0
Added missing mount command
commit sha f8c950c10828d4884070e3a1cdfd127858cffe76
Drop go 1.7 from test matrix, add 1.12 The current version of gensupport requires golang 1.8, so unless we vendor older versions building on go 1.7 will fail.
commit sha 4f31d5b9ae13e78b109002d55694bd3eeaf0dc81
Bump minimum CI golang version to 1.9
commit sha 3515c780d0a1f5154babfb838cbff12f1811861f
Test latest stable go instead of master
commit sha 9ada2b212c7c245661ca145060aecbb639a3ae3f
Merge pull request #321 from felixbuenemann/update-build-matrix Update build matrix
commit sha 4e954a0e7a0aa098a241c2df9eb02d875551cdeb
Add Team Drive Support (#319)
commit sha bf61a476178d450e969bbf8fcc910c4c4c2aa25b
README: Add how to get a Team Drive Root Node ID
commit sha c02407ed7dad52e1cc354a84aedbf07ee798416f
Remove allow_root mount option This is no longer supported, because fuse.AllowRoot() was removed: https://github.com/bazil/fuse/commit/e0b89709bc6481f1345b6e19363cf18766e53fc7
commit sha f5837b6facaa0f427e8be04ceb1a96e565a0c66a
Bump CI to at least Go 1.13 This is needed to build against the latest version of basil.org/fuse.
commit sha 6a8661b3a2248791496ad92229af27dd6fc91056
Merge pull request #340 from felixbuenemann/remove-allow-root-mount-option Remove allow_root mount option
commit sha f12bbe593d6d07ecbfd050596076109b7610cdfd
Add --drive-id option for team drives This fixes the problem where --root-node-id was broken for non-team drives and also should allow mounting a sub.directory of a team drive by specifying both --drive-id and --root-node-id.
commit sha 1486bf3d53de9e708c973ea6e2a54b8892b9242d
Ignore changes with changeType drive This should fix the unexpected change records without a file resource. Additionally we request to include the file resource for removals, which could be another reason of missing file resource in changes.
commit sha d2a14ee3bb27b5c9049ddfa795be7088acfa51cd
Update API to support all shared drives
commit sha afa5e4e8c125e2758398dc72ab19dc7ea74e1a8c
Merge pull request #342 from felixbuenemann/team-drive-fixes Team Drive Fixes
commit sha e205113c294b8cc4823d6a9d4818bfe816883e00
Remove allow_root option from README This option was removed in #340.
commit sha 7f5abf6fe0eb6c11bd1a703594af7280ccba130e
Fix off-by one error in download range request
commit sha 65ff5d41f985df55b3d30bb1bd65183626472c84
Prepare 5.1.0 release
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha e205113c294b8cc4823d6a9d4818bfe816883e00
Remove allow_root option from README This option was removed in #340.
commit sha 7f5abf6fe0eb6c11bd1a703594af7280ccba130e
Fix off-by one error in download range request
commit sha 65ff5d41f985df55b3d30bb1bd65183626472c84
Prepare 5.1.0 release
commit sha b6c8f1709a22cb889fb5975a754abd0bcf93d5ff
Add GitHub Actions Go Workflow (#378) This builds plexdrive and runs the tests.
commit sha 6baff616561d1e2627b7084e346d139731fd7910
Update all github repo references (#377)
commit sha 512522146aad6c2964cc6e1833f11d645957eb40
Fix reads across chunk boundaries When using Direct-IO requests are not guaranteed to be aligned on page size boundary, like with buffered IO. This means a read request can start at the end of one chunk and finish at the beginning of the next chunk. In cases where this happens we now request multiple chunks in GetChunk and assemble the response data.
commit sha e3389843ecec9b839f4507ca209931faa0d5e518
Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days
push eventfelixbuenemann/plexdrive
commit sha 86e7ac906c117f1a73418edef5c156c5c1053b81
Assemble request slices out of order This improved performance, since we don't have to wait for each slice sequentially, but can request them all at once. Since the channel order does not have to match the request order, we track a sequence number with each request to allow reassembly of the response out of order.
push time in 21 days