CachePerfStats gathers performance data for single open, read and write operations as well as the whole cache entry opening. It maintains long term and short term average. The long term average filters out excessive values and it represents and average time for a given operation when the cache is not busy. The short term average represents the current cache speed. By comparing these two stats we know pretty quickly that the cache is getting slower and then we race the cache with network immediately without a delay. Otherwise the delay is based on the average cache entry open time.
This patch makes the following changes to the macros.
- Removes PROFILER_LABEL_FUNC. It's only suitable for use in functions outside
classes, due to PROFILER_FUNCTION_NAME not getting class names, and it was
mostly misused.
- Removes PROFILER_FUNCTION_NAME. It's no longer used, and __func__ is
universally available now anyway.
- Combines the first two string literal arguments of PROFILER_LABEL and
PROFILER_LABEL_DYNAMIC into a single argument. There was no good reason for
them to be separate, and it forced a '::' in the label, which isn't always
appropriate. Also, the meaning of the "name_space" argument was interpreted
in an interesting variety of ways.
- Adds an "AUTO_" prefix to PROFILER_LABEL and PROFILER_LABEL_DYNAMIC, to make
it clearer they construct RAII objects rather than just being function calls.
(I myself have screwed up the scoping because of this in the past.)
- Fills in the 'js::ProfileEntry::Category::' qualifier within the macro, so
the caller doesn't need to. This makes a *lot* more of the uses fit onto a
single line.
The patch also makes the following changes to the macro uses (beyond those
required by the changes described above).
- Fixes a bunch of labels that had gotten out of sync with the name of the
class and/or function that encloses them.
- Removes a useless PROFILER_LABEL use within a trivial scope in
EventStateManager::DispatchMouseOrPointerEvent(). It clearly wasn't serving
any useful purpose. It also serves as extra evidence that the AUTO_ prefix is
a good idea.
- Tweaks DecodePool::SyncRunIf{Preferred,Possible} so that the labelling is
done within them, instead of at their callsites, because that's a more
standard way of doing things.
--HG--
extra : rebase_source : 318d1bc6fc1425a94aacbf489dd46e4f83211de4
All the instances are converted as follows.
- nsSubstring --> nsAString
- nsCSubstring --> nsACString
--HG--
extra : rebase_source : cfd2238c52e3cb4d13e3bd5ddb80ba6584ab6d91
All the instances are converted as follows.
- nsAFlatString --> nsString
- nsAFlatCString --> nsCString
--HG--
extra : rebase_source : b37350642c58a85a08363df2e7c610873faa6e41
All the instances are converted as follows.
- nsASingleFragmentString --> nsAString
- nsASingleFragmentCString --> nsACString
--HG--
extra : rebase_source : e6a1ddc8938fecd9a735b15e872c054edf4c7910
CachePerfStats gathers performance data for single open, read and write operations as well as the whole cache entry opening. It maintains long term and short term average. The long term average filters out excessive values and it represents and average time for a given operation when the cache is not busy. The short term average represents the current cache speed. By comparing these two stats we know pretty quickly that the cache is getting slower and then we race the cache with network immediately without a delay. Otherwise the delay is based on the average cache entry open time.
CachePerfStats gathers performance data for single open, read and write operations as well as the whole cache entry opening. It maintains long term and short term average. The long term average filters out excessive values and it represents and average time for a given operation when the cache is not busy. The short term average represents the current cache speed. By comparing these two stats we know pretty quickly that the cache is getting slower and then we race the cache with network immediately without a delay. Otherwise the delay is based on the average cache entry open time.
Adds nsISubstitutionObserver so that substitutions set on a
parent nsISubstitutingProtocolHandler which are then propagated
to child processes can be observed in the child.
Updates test_extensionURL.html to set substitutions on the parent
ExtensionProtocolHandler before trying to load moz-extension URI's
using those substitutions.
MozReview-Commit-ID: JaW1A3uZpoO
--HG--
extra : rebase_source : ca1b89b9a6ea29ea464979c26c8c48c61d9d7e00
Changes ExtensionProtocolHandler to use remote streams for
moz-extension loads of file and JAR URI's to allow for
filesystem read-access sandboxing.
Adds messaging to PNecko to allow child processes to
request an input stream or file descriptor for moz-extension
URI's.
Add ExtensionProtocolHandler singleton so that NeckoParent can
call methods directly and ExtensionProtocolHandler::NewFD
can use a new member variable |mFileOpenerThread| to open files.
Adds FileDescriptorFile, a limited implementation of nsIFile
that wraps a file descriptor, to be sideloaded into
nsJARChannels so that extension JAR files can be read
using a file descriptor without accessing the filesystem
directly.
MozReview-Commit-ID: 1pcnIpjz2yR
--HG--
extra : rebase_source : c3f2ede3dbfb29d98af6b194dbe58669d93d4197
SpecialPowers.loadChromeScript() sends a script to the child process,
then creates a sandbox, and runs the script in that sandbox. There are
various sandboxOptions that can be passed when creating a sandbox, and
it would be nice to have that functionality for loadChromeScript.
I just need this for wantGlobalProperties, but I might as well make it
as general as possible. I'm not sure all of the types it can take can
actually be serialized across processes, but I guess that's okay.
MozReview-Commit-ID: GoJjXdjizFk
--HG--
extra : rebase_source : 9c2bc190dbf5a080978953cffd64205e8b816367
Evaluated every method exposed via nsICookieManager and
nsICookieManager2, and ensured that every method uses the right cookie
database depending on whether the mPrivateBrowsingId flag is set in
OriginAttributes.
The usePrivateMode method has been removed since it was introduced in
bugzil.la/777620 and not used anywhere else.
This should be covered by
toolkit/components/extensions/test/mochitest/test_ext_cookies.html,
but due to bugzil.la/1309637, the relevant private browsing tests
have been disabled. These tests will be enabled in one of the next
commits.
MozReview-Commit-ID: DMhE42LGu9S
--HG--
extra : rebase_source : 7882ed7610ff4c7c6f81a7852660d2b36a5c0a92
Move the initialization of nsUrlClassifierDBService to prior of calling ShouldEnableTrackingProtection(), so the nsUrlClassifierDBService will be initialized before staring HTTP connection. If not doing so, there is a 3.x% peformance regression on android platforam.
Changes to Promise tests designed to test .then(null) have been reverted, and the browser/extensions directory was excluded because the projects it contains have a separate process for accepting changes.
MozReview-Commit-ID: 1buqgX1EP4P
--HG--
extra : rebase_source : 3a9ea310d3e4a8642aabbc10636c04bfe2e77070
As part of network security, we currently crash the main process in NeckoParent::GetValidatedOriginAttributes if the SerializedLoadContext is null. This makes it difficult to find out who created the channel that has a null SLC. Adding this assertion to the child allows us to get a nice stack trace and maybe identify the root cause.
MozReview-Commit-ID: 7rz91oEaiRi
--HG--
extra : rebase_source : b411e1e9494623bacb506a2f820902a3f7aa7c2a
Add a field to the HSTS cache which indicates the source of the HSTS
entry if known, from the preload list, organically seen header, or HSTS
priming, or unknown otherwise. Also adds telemetry to collect the source
when upgrading in NS_ShouldSecureUpgrade.
MozReview-Commit-ID: 3IwyYe3Cn73
--HG--
extra : rebase_source : 9b8daac3aa02bd7a1b4285fb1e5731a817a76b7f
As a follow-up from bug 1206961, we will remove calling CanLoadImage in
this bug. Also in the case of CSP check failed, we will call
SetBlockedRequest in those cases.
See https://bugzilla.mozilla.org/show_bug.cgi?id=1267075#c30 for the
analysis between the old and new setup.
Setting mFirstResponseSource in nsHttpChannel::OnStartRequest when cache wins is too late, because the channel might already started redirecting in nsHttpChannel::ReadFromCache. In this case nsHttpChannel::OnStartRequest is not call at all for the cache pump and we would proceed the network response which should be ignored.
Since the uri classifier needs topWindowURI to decide whether or not to enable channel annotation, we have to allow to change this attribute in js for passing the test.
--HG--
extra : rebase_source : c5effa05fecef0d32600e4c9c926dbfa77c2ca6f
This patch includes two changes:
1. The current channel annotation is happened at nsChannelClassifier::OnClassifyComplete and this is too late because this channel might be already hit the network.
This patch adds a new API CheckIsTrackerWithLocalTable in nsChannelClassifier to check if the URI is in local blacklist and whitelist before calling BeginConnectActual.
Please note that whitelist will be checked only when TP is disabled.
2. Skip channel annotation for first-party and top-level load http channels
--HG--
extra : rebase_source : 78f98f8200d3dbb1cb96cd0a33901b2a82227974
This is a preexisting issue that makes nsMultiplexInputStream multiple-inherit
from nsIInputStream: once via nsIMultipartInputStream and once via
nsIAsyncInputStream. This causes problems once we end up with more multiplex
streams that are async streams, because then some assingments to
nsCOMPtr<nsIInputStream> start asserting. This patch just removes the footgun
by getting rid of the multiple inheritance.
The original code creates nsChannelClassifier and calls ShouldEnableTrackingProtection twice when TP is enabled. To avoid redundancy, this patch makes channel classifier as a data member in nsHttpChannel. Note that the data member is a weak ptr to prevent ref count cycle.
--HG--
extra : rebase_source : 11d2e17c2b45b70e3cd407d882ddf0c57ad51755
This patch contains two parts below:
1. mNeckoTarget
Since almost all child channels in necko need mNeckoTarget to dispatch runnables to main thread, it seems worth to have a holder class of mNeckoTarget and ask child channels to inherit it.
2. ChannelEvent
A lot of classes in FTPChannelChild and HttpChannelChild inheriting ChannelEvent and override GetEventTarget. It should be worth to extract the same code and put it in the same space.
Collect telemetry for all requests to get an exact percentage of
requests that are subject to HSTS priming, and how many result in an
HSTS Priming request being sent. Clean up telemetry to remove instances
of double counting requests if a priming request was sent.
HSTSPrimingListener::ReportTiming was using mCallback to calculate
timing telemetry, but we were calling swap() on the nsCOMPtr. Give it an
explicit argument for the callback.
Add tests for telemetry values to all of the HSTS priming tests. This
tests for the minimum as telemetry may be gathered on background or
other requests.
MozReview-Commit-ID: 5V2Nf0Ugc3r
--HG--
extra : rebase_source : daa357219a77d912a78b95a703430f39d884c6ab
The current channel annotation is happened at nsChannelClassifier::OnClassifyComplete and this is too late because this channel might be already hit the network.
This patch adds a new API CheckIsTrackerWithLocalTable in nsChannelClassifier to check if the URI is in local blacklist and whitelist before calling BeginConnectActual.
Please note that whitelist will be checked only when TP is disabled.
--HG--
extra : rebase_source : 0761f6c7631bc934691c8d018be88514568a3aa1
The original code creates nsChannelClassifier and calls ShouldEnableTrackingProtection twice when TP is enabled. To avoid redundancy, this patch makes channel classifier as a data member in nsHttpChannel. Note that the data member is a weak ptr to prevent ref count cycle.
--HG--
extra : rebase_source : 4a77f1e51b38e27a065162cc702091aca2db51de
Also removes some dead code.
A lot of the code in ExtensionUtils.jsm is not needed in all processes, and a
lot of the rest isn't needed until extension code runs. Most of it winds up
being loaded into all processes way earlier than necessary.
MozReview-Commit-ID: CMRjCPOjRF2
--HG--
extra : rebase_source : 37718eaf05a22b8ccb95f633cf7454bd7975cdce
This removes unnecessary COM overhead from the extension protocol service,
particularly from the flag lookup code, which is called often, and from hot
paths. The devirtualized lookups should have virtually no overhead for
extensions without web-accessible resources, and very little overhead except
when resources are specified as non-prefix globs.
MozReview-Commit-ID: 4hQ7GuQSjvW
--HG--
extra : rebase_source : 61897a204abd915ad61852fa475cde2de51753f3
NotifyTrackingProtectionDisabled/NotifyTrackingResource/SetClassifierMatchedInfo updates tracking protection and safe browsing states from parent to child.
They should at least handled by child process before OnStopRequest. Move them to PHttpBackgroundChannel can ensure that.
MozReview-Commit-ID: 8T6k9z9FgrI
--HG--
extra : rebase_source : fd26e2b4309a31e0a9bb16c9236eb4a1875518dc
OnStartRequestSent is used to ensure the execution order between OnStartRequest and all other messsages sent over PBackground.
IPC messages sent before OnStartRequestSent can be processed immediately while received by content process.
IPC messages sent after OnSartRequestSent need to be pending until OnStartRequestReceived is called.
MozReview-Commit-ID: 5Vbb8SBQmW8
--HG--
extra : rebase_source : 81ddce949d2a658e227e1e67d7d4313f7c66513f
PHttpBackgroundChannel is created by content process because PBackground IPDL
can only be initiated from content process. The background channel deletion is
controlled by chrome process while PHttpChannel is going to be closed or canceled.
BackgroundChannelRegistrar is introduced for pairing HttpChannelParent and
HttpBackgroundChannelParent since they are created over different IPDL
asynchronously.
nsIParentRedirectingChannel.continueVerification is introduced to asynchronously
wait for background channel to be established on the new channel during the
Redirect2Verify phase.
MozReview-Commit-ID: 41l8ivan8iA
--HG--
extra : rebase_source : b8b6d7e7c037efaa9fc13df14191205c603e833a
AsyncOpen procedure can failed and trigger FailedAsyncOpen IPC to do the clean up.
However FailedAsyncOpen might not complete if content process is destroyed at the meantime.
We can delay the timing of holding the strong reference to parent listener and channel object
to make sure no reference cycle is created by HttpChannelParent. In addition, clean up the
strong reference as soon as FailedAsyncOpen IPC is triggered.
MozReview-Commit-ID: LDOt0BpBgFe
--HG--
extra : rebase_source : 38e1128e8c361a2422ec4d8292b591ab3b3ef41f
PHttpBackgroundChannel is introduced to deliver OnStopRequest/OnTransportAndData/OnProgress/OnStatus on background thread.
FlushedForDiversion/DivertMessages is moved to PHttpBackgroundChannel to ensure no more channel event are delivered after these two messages on the background channel.
OnStartRequestSent is introduced to synchronize the execution order between OnStartRequest (main thread IPC) and all other messages on PHttpBackgroundChannel.
MozReview-Commit-ID: HfqvPML6EKw
--HG--
extra : rebase_source : 423709763e1abf1362347bc2d3141e2d33aed46a
Avoids a thread-safety race condition on shutdown where we could try to finalize a statement twice.
Allows the async thread to be referenced until ShutdownAsyncThread, so async finalizers can make use of it.
Removes the no more useful mAsyncExecutionThreadIsAlive.
Nullifies the sqlite3_mutex pointer when the connection is closed, since it would be a dangling pointer.
Use a ScopeExit to ensure the connection and the mutex pointers are always nullified on failure.
Makes asyncClose bailout early if a Close method was already invoked before.
Makes AsyncInitDatabase not use AsyncClose to just shutdown the async thread.
Fixes various unfinalized statements in consumers code.
Makes mConnectionClosed better follow mDBConn status.
Replaces some mutex locking isClosed() calls with lockless isConnectionReadyOnThisThread.
MozReview-Commit-ID: 6sftFehsQTt
--HG--
extra : rebase_source : 2bf3088b500376e58e62e8f078d9950588adc649
This patch makes navigator object to return spoofed value for fields have fingerprintable
concerns. This changes the worker navigator as well.
MozReview-Commit-ID: E2SLNZRfuVP
--HG--
extra : rebase_source : 960bafbe41382eba891ad807c37a822ba6a758c8
This patch makes nsHttpHandler to use a spoofed userAgent in the User-Agent header
when 'privacy.resistFingerprinting' is true, the spoofed userAgent string has a
fixed appName, platform and buildID. The only variant in it is the version number,
it will be rounded to the nearest 10(e.g. 40, 50, 60) for reducing fingerprinting
across release versions.
MozReview-Commit-ID: BTGuIJp94U1
--HG--
extra : rebase_source : ab0b6379630c0e962636e624f2caccca6cc9f934
When the last request is removed from the load group, we report telemetry for the default load request. This was done without checking if the request was successful, which may cause us to report telemetry for failed requests as well.
Also, the NullHttpChannel had its timingEnabled attribute set to true, which could lead us to report invalid telemetry
MozReview-Commit-ID: 5w7rd2V17Xd
--HG--
extra : rebase_source : 60785ebc38da8880aa6ded668fed8af81c3d60e9
We write a lot of 4-bytes prefixes to file which call many system calls.
We should use a buffer and only write to file if the buffer is full or
finish writing. nsIBufferedOutputStream is a good candidate to do that
MozReview-Commit-ID: CzGOd7iXVTv
--HG--
extra : rebase_source : 25f1ce804b9a53e0a0a4023a1aa91f1a0ed98547
In order to provide more details context of how client arrived at the unsafe
page, particularly in redirect case, we may have to add more information to
redirect chains including:
- referrer (if any)
- remote address.
- URL
We may want to use an idl interface instead of nsIPrincipal to store these
information
MozReview-Commit-ID: 3Uh4r06w60C
For a long time, opening a resource:// url that leads to a file list has
used a title of the form "Index of jar:file://..." where the
jar:file://... url is the actual location the resource:// url has been
resolved to in the omni.ja.
That same url is used to derive a link to the parent directory. Because
of security context restrictions, the resource://... page can't open a
link to jar:file://... .
So we use the original resource:// url to derive the parent directory
link, and while here, also fix the title.
--HG--
extra : rebase_source : f12ff784c1671f15b83ff082eb7418fcb71f1e10
1. Use ScopeExit to ensure mListener->OnStartRequest() is invoked before exit CallOnStartRequest.
2. Add assertion to ensure OnStartRequest called before OnStopRequest.
MozReview-Commit-ID: FgONlk5HPNz
--HG--
extra : rebase_source : ffe386bc8bea6defb216ecf149a29deacdb3737a
See bug 1365483 for the rationale.
MozReview-Commit-ID: 4sWodnP5MYZ
--HG--
extra : rebase_source : 6a32073e50695ac8f4a540c83f9d2b5542b53585
extra : intermediate-source : 5d7189453d82844d55e3d90c79792b2edd7bdd0d
extra : source : 8b029c32f3512c9d8e92582312459a501594a6e8
This function is arguably nicer than calling NS_ProcessNextEvent
manually, is slightly more efficient, and will enable better auditing
for NS_ProcessNextEvent when we do Quantum DOM scheduling changes.
If HttpChannelChild::Cancel is triggered off main thread while flushing channel event queue,
CompleteResume might be executed while the flush is about to be finished. In this case, queue resumption
will not be able to trigger the second queue flush because the previous one is not finished yet.
Therefore, the HttpChannelChild::Cancel will be sitted in the queue without executing.
MozReview-Commit-ID: GxnkiDUmEnw
--HG--
extra : rebase_source : d04268a04ab24c38bf6fa436526a04f77cd64805
nsAHttpTransaction::Available() obtained a bytecount from the abstract
transaction's input stream. If that stream was derived from a file://
it would create janky IO - so remove the interface.
Http2Push maintains a non-inherited interface which is used to check
the number of bytes it has internally buffered in memory.
MozReview-Commit-ID: IQHt8yGsqDE
--HG--
extra : rebase_source : 64449c6bd743119ea7626a3b2b2b91a376280021
UAOverridesBootstrapper.js is introduced to delay the initialization of
UserAgentOverrides.jsm until the creation of nsHttpHandler in chrome process.
Uninit will be triggered at profile-change-net-teardown because no network
traffice after this point.
MozReview-Commit-ID: F8Lpn6RyZEm
--HG--
extra : rebase_source : b516209f96ec81deb54aab3c038803beb3cea441
system-info might need to be construct while creating nsHttpHandler and it might take up to 30ms.
Lazy loading the DEFAULT_UA can delay the creation of nsHttpHandler after start-up.
MozReview-Commit-ID: FtIpKjcY38r
--HG--
extra : rebase_source : 8061ed3ce6c42955e52f494166958f5b63ab940b
This is also the non-broken way to fix bug 1346392. Instead of waiting
until the auth handler gets its hands on things, we break layering a bit
and inspect the response headers as soon as we decompress them to see if
there's any connection-oriented auth being requested. If there is, we
treat the situation as if we got a RST_STREAM or GOAWAY with
HTTP_1_1_REQUIRED.
We were able to re-purpose the NS_ERROR_ABORT code path that was
previously used with an inappropriate HTTP status code when talking to
an HTTPS proxy over http/2, as that usage was removed a while back from
the stream, though we still had the (dead) code in the session to handle
the stream giving us that return value. The error code was changed to
NS_ERROR_NET_RESET, however, to give a better description of what's
going on.
MozReview-Commit-ID: DLjOIIiXGrV
--HG--
extra : rebase_source : 703fde39432808cabd05b48aa40165e53ebc5ed1
UAOverridesBootstrapper.js is introduced to delay the initialization of
UserAgentOverrides.jsm until the creation of the first nsHttpChannel.
Uninit will be triggered at profile-change-net-teardown because no network
traffice after this point.
MozReview-Commit-ID: F8Lpn6RyZEm
--HG--
extra : rebase_source : 7c3649b50ad8594dc0968961fbbd2766d0d98b0a
system-info might need to be construct while creating nsHttpHandler and it might take up to 30ms.
Lazy loading the DEFAULT_UA can delay the creation of nsHttpHandler after start-up.
MozReview-Commit-ID: FtIpKjcY38r
--HG--
extra : rebase_source : 8061ed3ce6c42955e52f494166958f5b63ab940b
nsAHttpTransaction::Available() obtained a bytecount from the abstract
transaction's input stream. If that stream was derived from a file://
it would create janky IO - so remove the interface.
Http2Push maintains a non-inherited interface which is used to check
the number of bytes it has internally buffered in memory.
MozReview-Commit-ID: IQHt8yGsqDE
--HG--
extra : rebase_source : 78dbd5cae35bc6cb1ce2f03192226cb85564298e
nsAHttpTransaction::Available() obtained a bytecount from the abstract
transaction's input stream. If that stream was derived from a file://
it would create janky IO - so remove the interface.
Http2Push maintains a non-inherited interface which is used to check
the number of bytes it has internally buffered in memory.
MozReview-Commit-ID: IQHt8yGsqDE
--HG--
extra : rebase_source : 6b3b210a8f6736887b5076c90621d678b038bcad
This patch prevents the following error:
AltDataOutputStreamChild::Close -> SendClose()
AltDataOutputStreamChild::Release -> Send__delete__()
AltDataOutputStreamParent::RecvClose() -> SendError()
AltDataOutputStreamParent::ActorDestroy -> mIPCOpen = false
AltDataOutputStreamChild::RecvError -> === Crash - object was deleted ===
We introduce the DeleteSelf message.
AltDataOutputStreamChild::Release -> SendDeleteSelf()
AltDataOutputStreamParent::RecvDeleteSelf -> mIPCOpen = false; SendDeleteSelf()
AltDataOutputStreamChild::RecvDeleteSelf -> Send__delete__()
The parent will not send any more messages after receiving the DeleteSelf message.
MozReview-Commit-ID: I9RQ5I7eSt9
--HG--
extra : rebase_source : 8f918d24595248149ebd3857e05e38dc5237059b
We now have code that unconditionally requires the rust
compiler and are committed to adding more. Remove this
last vestige of conditional support.
MozReview-Commit-ID: EK6FBnAbR
--HG--
extra : rebase_source : 6efda10a74f9ca0482304c2b1ffe6941e42138f8
The channel objects cannot be handed off to other threads before the creation
process has been finished, so there is no point in trying to hold these locks
while the initialization code is running. These lockings have shown up in
profiles as being expensive.
Michal Novotny
Nicholas Nethercote
Phil Ringnalda
Dragana Damjanovic
Sebastian Hengst
Carsten "Tomcat" Book
Junior Hsu
Kershaw Chang
Wes Kocher
Haik Aftandilian
Andrew McCreight
Florian Quèze
Valentin Gosu
Kate McKinley
Shih-Chiang Chien
Liang-Heng Chen
Rob Wu
Paolo Amadini
Masatoshi Kimura
Ben
Honza Bambas
Jan Keromnes
Bill McCloskey
Yoshi Huang
Ben Kelly
Mats Palmgren
Boris Zbarsky
Jan de Mooij
Tom Ritter
Ryan VanderMeulen
Nicolas B. Pierron
Tom Tromey
Andreas Farre
Ehsan Akhgari
Kris Maglione
Randell Jesup
Nils Ohlmeier [:drno]
Randall Barker
Bevis Tseng
Marco Bonardo
Tim Huang
Henri Sivonen
Thomas Nguyen
Arthur Edelstein
Mike Hommey
Jonathan Kew
Felipe Gomes
Daniel Stenberg
JW Wang
Will Wang
Dragana Damjanovic dd.mozilla@gmail.com
Edouard Oger
Michael Kaply
Iris Hsiao
Catalin Badea
Christoph Kerschbaumer
Nathan Froyd
Patrick McManus
Nicholas Hurley
Milan Sreckovic
Ralph Giles
Andi-Bogdan Postelnicu
Michael Layzell
Jonathan Kingston