Less tracking of offsets and also more similar to how
firmware does this.
We rewrite our current position on tool change to current position
plus tool offset instead of applying it for every single
position change.
We fetch settings once explicitly after passive login but before
completed startup. Without this patch we fetched them again after
onUserLoggedIn got fired (if it got fired by the passive login)
during startup. That's not necessary because we did the passive login
before our initial settings fetch, so IF that already logged us in,
our settings fetch already was done with us logged in as well and
the onUserLoggedIn later only tells us what we already knew.
We still had some concurrent requests triggered through onStartup that
managed to overtake the passive login (and hence nuke the session
data on the client). This should now really not be possible any longer.
Also created a sequence diagram of the new process and linked it from
a lengthy comment.
It caused issues with our line number tracking in the browser
developer tools and I don't see any way to avoid this considering
that we can't adjust the arguments supplied to console.log without in
fact wrapping it :/
Under specific circumstances it could happen that the passive login
response was overtaken by a response from an earlier still
unauthenticated request which would then effectively overwrite the
session cookie immediately and hence cause the browser to use its
login session against the server, but without displaying that to the
user. See #1881 for what is probably an issue caused by exactly this
kind of scenario.
Additionally a couple of requests needed to be done a second time
after login, just moments after having done them for an anonymous
user. See #1572 for a request to change this behaviour.
This patch changes the startup order of the web interface like this:
1. connect to the websocket, postpone any events triggered by
this (e.g. "fromHistoryData") until 3
2. perform a passive login, postpone any events triggered by this
(e.g. "onUserLoggedIn", "onUserLoggedOut") until 4
3. trigger postponed events from 1
4. trigger "onStartup" (triggers postponed events from 2)
5. fetch settings
6. bind view models etc
A connection close of the websocket will disable event processing on the
socket until it is marked as initialized again by the passive login
processing, which will also be triggered immediately on server connect.
That way under normal circumstances nothing should ever get triggered
in the registered view models that might generated requests to the
server API until a passive login has been done, a server session
initialized and if necessary a user properly authenticated.
They might be nozzle priming routines which are not actually part of
the model proper and hence it doesn't make sense trying to keep them
visible when zooming in on the model, or allowing them to offset the
model center.
Now also detects capabilities reported by (extended) M115 output and
if AUTOREPORT_TEMP is available enables it with the configured
autoreport interval and disables active polling.
Implements #1679
See #1942
This does NOT fix the root cause of `_default.profile` becoming an
empty file (apparently during migration from config.yaml?), however
since this particular situation was impossible to reproduce so far this
workaround for the consequences will have to suffice for now since we
want a hotfix as fast as possible for 1.3.3.
* fa-circle-blank does not exist, replaced with fa-circle-o (see also
report in #1928)
* fa-trash-o instead of fa-trash (same look and feel)
* fa-check instead of fa-check-circle (same look and feel)
* fixed toggleContent binding to also be able to toggle fa icons
* fix deletion spinner in file list (was still using old icon-
classes which of course weren't there anymore and hence not
toggleable)
* icons-ul => fa-ul
* remove superfluous fa class from some places
Follow-up to #1915.
Instead of disabling capturing of postroll by default (which we actually
don't want and doing so was a mistake thanks to misremembering the
meaning of the variable in question), we now properly reset the
default value for that check box (which wasn't properly set only due to
a very stupid typo).
That fixes up the relative URLs to the font files and should make updates down the line
a thing of a quick copy and paste (plus allow some more sophisticated dependency
management down the road as well).
Also renamed fonts folder for FontAwesome 3.2.1 from font to fonts, for more
consistency.
Otherwise we will block ourselves, waiting for the restart command to
complete which it only can when we are no longer there.
Should reduce restart times on update significantly.
Downside is that we no longer can wait for the return code of the
call. However, that should be caught by our UI handler timing out for
the restart and showing an error prompting the user to restart
manually.
By default G90/G91 define relativeMode, M82/M83 define relativeE. The
extruder coordinate will be viewed as relative when either
relativeMode or relativeE are set. This mirrors the implementation in
Marlin (current and legacy forks) as well as Repetier.
SmoothieWare's implementation resets the relative mode of the
extruder on a G90/G91. To still accomodate this different
implementation a new settings has been added under Settings >
Feature: "G90/G91 overrides relative extruder mode". Checking this
switches to the alternative interpretation for future gcode analysis
and visualization.
The only difference between the two of them is that the following
segment of GCODE:
M83
G90
will have the extruder still set to relative on Marlin and
Repetier but set to absolute on Smoothieware.
Please note that the setting in all likelihood will be moved to
the printer profile in the future, since it makes more sense to have
it on a per printer basis. It's not part of the set of features
available for auto detection since it's required for GCODE analysis
and hence needs to be known even outside of a connection to the
printer.
Implements #1818
An aspect ratio of 16:9 is assumed, with other ratios showing black
letterbox borders as required.
During loading a text "Webcam loading..." is displayed in the webcam
space. On loading error, the (broken) image is hidden and an error
text is displayed instead.
Solves #478
Otherwise we might run into a rare race condition where the M110 sent
on start of a SD card upload gets prepared for sending after the
source file is already opened. That would then lead to the M110 not
getting processed anymore by OctoPrint and hence the line number
counter not resetting accordingly, leading to one line number
mismatch after the next from the firmware.
Testing if our "isPrinting" flag is ALSO set to true here ensures
that we'll only stop processing commands internally once the state
has actually switched to printing, which only happens after the
firmware responds to the M28 sent after the M110.
Fixes#1882 and probably also the issue encountered by @amd989 in
#1762 (which looks like exactly the same problem)
Disable button when a test is already active. Also have spinner
defined centrally and only made visible on active test.
Should hopefully give better feedback to impatient users ;)
Solves #1897
* Use click data-bind for submit button instead of manual click
binding
* Keep metadata synchronized (otherwise changing identifier, name,
description, overwrite or default flag after adding a file does
absolutely nothing)
* Disable submit button and fields when no file is selected.
* Disable submit button (and show spinned) when request is in
progress.
Related to #1893
This fixes#1893.
Also, only bind click to the "Confirm" button once instead of binding
and unbinding it for each file. Binding it just once seems cleaner.
(cherry picked from commit 83d1921)
If double the speed needed for output during print/fancy terminal
functionality is detected, re-enable that after a timeout of 5s.
Decrease of speed cancels the timeout again. That should prevent wild
toggling.
Should solve #1862
Some users get confused by having to use the blue "Print" button in
the state panel for consecutive prints of the same model, let's solve
this once and for all.
Move the AutoScroll function to an animation frame to prevent layout thrashing and to keep it in sync with fastForEach
Remove the <br> tag from the terminal and use display: block; on span tags to prevent increasing DOM nodes and faster processing of the terminal
On call of the lazyload plugin, all matched img elements will be
checked if they have a data-src attribute. If so, their src attribute
will be replaced with the contents of the data-src attribute
(effectively loading the image) and the data-src attribute will be
removed (so that multiple calls don't call multiple processing).
* Don't restrict the list of compatibility values to check against
to only those we have mapped, also support unmapped more exotic
ones.
* Allow 1:1 check against sys.platform values (with startswith).
Combined with the above that allows very granular compatibility
modelling ("freebsd11", "freebsd12") if required.
* Instead of only whitelisting ("linux", "freebsd") now also black
listing is possible ("!windows").
A detected os must match all provided whitelist elements (if the
whitelist is empty that is considered always the case) and none of
the backlist elements (if the blacklist is empty that is also
considered always the case).
See the included unit tests for examples of how this works.
Apparently python setup.py develop tries to re-install all scripts
provided by plugins (maybe only under Windows?). Since that includes
OctoPrint's own executable, things will explode with an access denied
error there. Switching over to pip install -e . fixes that, and also
is more conform with the install approach the plugin will take in
production anyhow.
So far using the "... from URL" or "... from an uploaded archive"
mechanism for installing a plugin would fail without further
information if the plugin to be installed was already installed.
The plugin manager will now detect this situation by parsing the
corresponding message from the pip output and trigger a reinstall
instantly.
A message about this will be logged to octoprint.log and the install
output.
Additionally the error handling for installation errors has been
slightly improved (install output no longer says "Done!" but "Error!"
with the reason as provided by the server) and the "could not install
plugin from URL unknown" issue should also be solved.
Evaluates json data file as e.g. published on
plugins.octoprint.org/notices.json and displays notices for plugins
installed and matching (optional) version lists.
* Extended plugin metadata by new property, only evaluated for bundled
plugins.
* Adjusted plugin manager to evaluate new metadata and add
confirmation dialog with details when attempting to disable such
a plugin.
* Added combined OctoBlog feed, replacing news + spotlight (+
octoprintonair), added corresponding config migration
* Subscribe to all registered feeds by default
* Added config button to announcement reader
* Added note how to edit announcement subscriptions to notifications
* Auto-hide announcements on logout
* Order channels server-side based on new order config setting
Old default behaviour of showing logged in users the update
notification stays default, but can now be disabled via the software
update plugin settings.
Additionally added the ignore button to the notification for users as
well (and made ignore entry in local storage user specific to still
show notification to other logged in users), plus a small hint that
in order to apply updates an admin is needed. Additionally now hiding
the notification on log out.
Closes#1739
Also introduced a new helper, a progress modal that can be used for
providing feedback about things such as bulk delete operations in the
background.
See #748 and discussion in #1807
It now resides in a file on disk and can be deleted. If the profile
configured as default is not available (e.g. due to having been
manually deleted, or on first start), the default profile as defined
in the profile manager will be written as _default and that will be
used/set as default.
On first start after this change gets applied, OctoPrint will migrate
the old "default overlay" residing in config.yaml to a file on disk
and delete the old entries in config.yaml.
Solves core reason of #1734
Addition of a notification under the save button when there are unsaved
changes, and a notice under the save as default checkbox when unticked,
that changes are only temporary.
Also move the "interval" box up the top to position it underneath the
timelapse type dropdown list, and change the text to be a bit more
descriptive.
(cherry picked from commit 381a016)
System wide API key now offers a generate button like the user
API keys. Setting the API key directly to a value via the settings API
endpoint is now no longer possible, which should prevent setting it
accidentally thanks to the browser prefilling things where it shouldn't.
No delete button is offered for the system wide API key since it will
get automatically regenerated on server start if not set, so
regeneration is the only functionality here that makes sense.
If no API key is set in the user settings, the "delete" button is now
disabled. If a key is already set and a new one is to be generated, a
confirmation dialog makes sure this is really what the user wants. Same
for deleting an existing API key.
Both the system wide API key and the user specific API keys will now
only display a QRCode if there's actually a value for the key.
Might confuse users to see that marked up like an error when it's
perfectly normal for some kinds of installs simply because the
specific build folders never were created in the first place.
Force --all on stock clean command unless --orig is specified, in
which case "python setup.py clean" behaves like the stock clean
command. Additionally there are now --noeggs to skip egg deletion and
--nopyc to skip pyc deletion.
Might have been an external reset of the printer we didn't otherwise
detect, and unless we reset the line numbers we'll now be off and run
into resend requests.
Solves #1676
* UserSettings: crashed if overwrite an existing key with primitive value by same key with complex value. Also reduced number of calls to user.get_setting()
* Updated tests
(cherry picked from commit b75b53d)
On reload of the user settings a new user object was created but not
updated on the associated session user(s). That let to session user
(s) still referring to the old user instances with the only partially
updated user settings.
Behaviour changed such that on a reload of the users from file
(including user settings) all registered session users will now get
their reference updated as well.
Also made modelling of user id to session id a bit less redundant.
Former implementation tested for z-hop rising flank and anything that
is falling. That however causes issues with z-hop directly followed
by layer change (diff is not z-hop size but negative, image hence not
captured).
Instead of testing for positive flank == z-hop and any negative flank,
we now test for absolute value == z-hop.
Solves #1777
Cache was not invalidated on change of:
* snapshot URL presence
* system menu entry presence
* gcode viewer enabled/disabled
* changes in access control availability
Refactored cache setup to add enable_timelapse, enable_gcodeviewer
and enable_accesscontrol to ETag calculation, removed system menu
check. Template filtering will now be done based on the values of
these flags at render time, not before.
See also #1776 for a related ticket.
* Use only icons for settings and system entries
* Limit length of login entry to 100px
* Limit length of instance name to 250px
* Removed "OctoPrint" text if instance name is set
See discussion in #1741
* leading and trailing null bytes will now be stripped
* while connecting, a "wait" will also trigger switch to connected
state - relevant when connecting without reset and initial "ok" for
M110 gets mangled
* extended virtual printer by new options to test the above:
* mangled "ok" can now be injected via "preparedOks" setting and
prepare_ok DEBUG command
* simulated reset lines on beginning ("start" etc) now only sent
if new setting "simulateReset" is true
Should fix#1770
Extracted methods to check if modification of a file is possible and
to determine if a specific file is currently select to take care of
denying upload and re-selecting file after upload.
Refactored watched folder handler to use the same methods and mirror
behaviour of file upload via API.
Should solve #1760
* Endless recursion on file delete as reported in #1738
* Broken implementation of .move
* Broken definition of .pathForEntry, .entryForPath,
.pathForElement and .elementByPath
API endpoints that were not decorated with restricted_access so far did
not properly create a user session for cookie-less requests with an API
key only.
That meant that flask.current_user stayed an anonymous user, even with
the global admin API key. In case of /api/settings that meant that even
with the global admin API key it was not possible to read settings that
are admin-only (like the API key for example) or user-only.
This has now been rectified by moving the session creation into a
different before_request handler that is registered globally on all API
endpoints, meaning that an API key will now always lead to a user
session to be created, regardless of the requirements of the API
endpoint in question. Additionally the CORS handling was extracted
as well as the API key presence enforcement.
BlueprintPlugins will now also get the CORS and session-from-API-Key
treatment if they do not declare their endpoints as restricted.
Might solve the API key "n/a" issue filed in #1751, but since the cause
of that isn't identified yet that's not sure.
