This slows things down terribly, particularly the setting on the websocket
made tiles appearing one by one. Let's keep the possibility to zero the buffer
sizes for debugging, but hide that behind an env. variable (and in debug
builds only anyway).
Change-Id: Ie4d2cdb3c0ec3c50f1a2b4f9941a462ac4f2d196
This is rare and should take a few short seconds
to save and then we can load as normal.
Since the client auto-reconnects anyway, no
need to tell the user to try again.
Change-Id: I51a7ad207edec3203933a1808288870e69699952
Reviewed-on: https://gerrit.libreoffice.org/36655
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
The server correctly saves all documents
and waits to upload them before exiting.
Change-Id: I04dc9ce588bc0fa39a9deb298d0a5efa61a03f1a
Reviewed-on: https://gerrit.libreoffice.org/36654
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
A Document has its own snapshots set.
A snapshot is a string representation of a JSON object.
AdminModel keeps also the expired document objects.
Query each document object in order to get their own history.
Admin accepts an "history" command then returns a json object.
An administrator checks the history by dashboard.
Change-Id: I73c87eff334cdb5a4a58043b2b66f18a56240b3a
Reviewed-on: https://gerrit.libreoffice.org/35926
Reviewed-by: pranavk <pranavk@collabora.co.uk>
Tested-by: pranavk <pranavk@collabora.co.uk>
When calling the cleanup from DocumentBroker::destroyIfLastEditor(), it might
be (theoretically) too early; so let's do it safer - as the last thing when
shutting down the DocumentBroker::pollThread().
Change-Id: Idc5e34b36d524c8f87a3a0fc99c87244e9784abf
Client now sends setclientpart only when it wants to change
the current part number, but not when it is updating
the current part when it receives setpart from the server.
When switching parts (either by the keyboard or mouse)
LoKit sends a notification of the current part and
invalidates tiles. As a reaction to that the client
cancels its tiles and requests new tiles for the
new part. In addition, it sends setclientpart
back with the new part number.
This works fine, except when the client switches
between parts quickly (as when randomly pressing
page-up and page-down quickly in impress), by
the time LoKit receives the second request the
client would have reacted to switching the first
by sending setclientpart for the first. This
will result in a resonance back and forth ad
infinitum (or when the client goes inactive).
The client shouldn't echo back the part number
which the LoKit has told it (it should only send it
to explicitly set it itself). But the server should
be smart enough to remove duplicate setpart
notifications from the client queue. This reduces
the chances of an ocillation like that to happen.
Change-Id: Ia8a1bb25e427ab910129340ce599e3743983463b
Reviewed-on: https://gerrit.libreoffice.org/36598
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
When switching parts (either by the keyboard or mouse)
LoKit sends a notification of the current part and
invalidates tiles. As a reaction to that the client
cancels its tiles and requests new tiles for the
new part.
Since the response to setpart is a large number
of tile cancellation and new requests, the
cost of setpart is significant (esp. with large
number of clients). To avoid that, we de-duplicate
setpart entries from the client queues so there
would be only at most one setpart (the last)
queued to be sent to the client. This minimizes
the unnecessary noise when the part changes
faster than the server and/or network can
keep up.
Change-Id: I87578004203acc63d43e6d398ca04e37f766d9ba
Reviewed-on: https://gerrit.libreoffice.org/36597
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
Retry up to 10 times with a delay of 1 second in between.
Also, show the user a message (after first reconnection
fails) that the document is unloading to try again
(but of course we auto-retry anyway).
Change-Id: I616d7643f146a6d5b7ccda7bb2aa567e3a9efe47
Reviewed-on: https://gerrit.libreoffice.org/36596
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
Without an explicit WS message, the client
does not get this message and the handler
is not invoked at all.
Change-Id: I71e210a9958965cff35dd4d0f1d99985429b82f4
Reviewed-on: https://gerrit.libreoffice.org/36593
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
This box-shadow interferes with direct comment children and breaks the
root-children comment block UI. Lets remove the shadow for now.
Change-Id: Idda659bcfca76093c148400b2ab311bc39861e84
Change the layouting algorithm to check for comment relationship and set
the layout bounds accordingly.
Change-Id: I67cc32092d8cf2c1bbc0d677258387739bfc638e
There is a race between creating and adding
a DocBroker into the DocBrokers container
and cleanupDocBrokers is invoked (on timer)
before it had a chance to start its poll
thread. This is exceedingly rare, but it
has happened.
We check that_stop==false flag when deciding
isAlive such that cleanDocBrokers will
not remove it before its thread had
a chance to run (which would happen
after adding it and creating the
ClientSession).
Also, no point in checking isAlive from
the polling thread itself (of course it is
alive).
Change-Id: If54fe2b5fce0697ee0e2f38f1662c71105e29347
Reviewed-on: https://gerrit.libreoffice.org/36500
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
And few other improvemnts inluding:
* use svg instead of png for the menu
* add title (caption) for the Accept / Reject change.
Change-Id: Ic7e781d7e93d319f766b387a8eddfa70c1920760
Some older browsers don't have meta tag support for CSP. Lets put all of
the CSP in response headers to be compatible with oldies.
Change-Id: I7f0d7c294e492b3c69ebea6fbd820d6558b9c3b3
This is actually not a displaced cursor, but displaced tiles map pane.
It happens when the user refreshes the the document page and
before the document finishes loading, switches to some other tab and
then get back to the document when the document load finishes. In such
circumstances, due to browsers not emitting the 'resize' event (probably
because it didn't have the focus when the map loaded) we return
incorrect/unexpected map center. Because 'resize' event sets this._initialCenter to
null, so map.getCenter() never returns this._initialCenter and instead
return this.layerPointToLatLng(this._getCenterLayerPoint()) which seems
to be the correct thing to return here.
The reason that the displaced cursor is not
observed when user doesn't switch to other tabs is because of the
browsers emitting the 'resize' event before we set the map transforms.
Nevertheless, in some circumstances it is quite possible that this
event, 'resize', is processed after we set the transforms even though
user hasn't switched the tabs but probability of this is very less which
justifies this bug's hard-to-reproduce nature when user doesn't change
the tabs.
Instead of making sure that 'resize' event is triggered before we set
the transforms, removing this block of code that returns unexpected
return value (which we never seem to use anywhere anyhow) seems more
sensible thing to do.
Change-Id: Iff532a902e6100be7f39c204cbf2f28f1a7f6a49
Reviewed-on: https://gerrit.libreoffice.org/36458
Reviewed-by: pranavk <pranavk@collabora.co.uk>
Tested-by: pranavk <pranavk@collabora.co.uk>
When we are just interested in equality. compare() is more meant for
sorting functions where negative/zero/positive return value is useful.
Change-Id: I11138a14dc08e23d33f3848aeb734d9f56f3e9f7
The number of outstanding child forks can
become negative if more children are
spawned than requested.
This prevents such a scenario from
permanently preventing WSD from spawning
new children, which happens when
OutstandingForks is negative.
Change-Id: Ief1e56d7b4a079e097ca2d18bd90a01d935f6b30
Reviewed-on: https://gerrit.libreoffice.org/36437
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
Commit 1e1f23716c fixed this already by
introducing by-value parameters, but
8a1f321c84 broke it. Fix this again, this
time more explicitly.
Change-Id: If29250ac2e99855796935b5cc05ccb222f8a4ad5
Reviewed-on: https://gerrit.libreoffice.org/36436
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Tested-by: Michael Meeks <michael.meeks@collabora.com>
I don't think we should leak our address
(which mostly is behind a WOPI host and end-user
has no idea of what host LibreOffice Online is running at) in the
Referer header. Lets be more strict here and don't leak our address
at all.
Change-Id: Ibc30e9b64e2e06e2e8d541c5f089320ecb11412b
Though this guard the user against MITM attacks, but enabling this also
has the potential to brick your websites. So, do not use it/enable it
without understanding what it actually is.
See: https://developer.mozilla.org/en-US/docs/Web/HTTP/Public_Key_Pinning
Though this should work, but I have not been able to test it because of
Firefox and Chrome's limitation/feature that key validation is not done
when certificate chain terminates at a user-defined trust anchor and I
couldn't find any way to temporarily enable the HPKP key validation for
such CA chains.
Change-Id: I64d4ff82b04c59642fa7b8bac2f8788a03950b28
Reviewed-on: https://gerrit.libreoffice.org/36357
Reviewed-by: pranavk <pranavk@collabora.co.uk>
Tested-by: pranavk <pranavk@collabora.co.uk>
This reverts commit de2bc17c04af088d9c7e18a97216b174494e1a9c.
Lets not introduce any cleanup commits while we are near a release, will
apply it again after the release. The cleanup is supposed to not handle
the custom file server root correctly, so don't forget to test it with
a custom file server root before re-reverting.
It changes the path where loleaflet.html is searched for from
/usr/share/loolwsd/loleaflet/... to /usr/share/loleaflet/...
and doesn't find it there.
Change-Id: I23940e9a3e06721f0a8b7493a526f42d2072cfa4
