chore: bump uvicorn to 0.25.0 #664
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
dhruvsyos
added a commit
to dhruvsyos/chainlit_sysos
that referenced
this pull request
Jul 15, 2024
* remove unused recoil dependency (Chainlit#551) * make sure context is always carried (Chainlit#552) * restore context * use a copy of metadata when restoring the user session to avoid conflicts and json serialization issues (Chainlit#557) * update langchain cache (Chainlit#556) * prepare 0.7.604 * fix lc placeholder with lcel (Chainlit#568) * call on_chat_end when session is manually cleared (Chainlit#567) * avoid readme flickering when resuming a chat (Chainlit#566) * avoid readme flickering when resuming a chat * highlight resumed conversation in the sidebar * redirect to main chat after deletion * avoid error if conversation does not exist * check module is loaded before deleting it (Chainlit#563) * check module is loaded before deleting it * fix key error * allow html behind feature flag (Chainlit#565) * allow html behind feature flag * make latex a feature * prepare 0.7.700 * fix chat_profiles + clear asks * feat(playground): add gpt-4-turbo in the playground (Chainlit#578) * Update README.md * Release/1.0.0 (Chainlit#587) * init * base data layer * add step to data layer * add queue until user message * remove data_persistence from config * upload askfilemessage response as file element * step context * step context * llama index integration + step elements * haystack integration + step error * langchain integration + error handling * feedback * feedback * refactor AppUser to User * migrate react-client * migrate react-components * migrate main ui * fix mypy * fix type import * fix step issues + langchain issues * token count * remove IMessage and MessageDict * wip fix tests * fix existing tests * add data layer test * action toast * remove seconds from message time * add support for action interruption * rename appuser to user * toast style * fix update thread * use http for file uploads * remove useless create_task * wip data layer * rename client step type * fix chainlit hello * wip data layer * fix test * wip data layer * add root param to step * fix llama index callback handler * add step show input * fix final answer streaming * update readme * step type lower case * chainlit_client * debug ci * debug ci * bump sdk version * bump version * bump versions (Chainlit#590) * various bug fixes (Chainlit#600) * various bug fixes * file upload should flush thread queues * relax python dependency * fix deps * don't send steps if hide_cot is true * make oauth cookie samesite policy configurable * add copy button and replace message buttons * 1.0.0rc2 * changelog * bump sdk version * add page to pdf element (Chainlit#606) * do not ask confirmation for new chat if no interaction happened (Chainlit#605) * do not ask confirmation for new chat if no interaction happened * fix test * Wd/fix element (Chainlit#608) * fix element url when authenticated * fix file url * fix chat resume first interaction * hide copy button is disable_feeback is true * add changelog * fix flaky tasklist test * feat(playground): Add Gemini Pro in the LLM playground (Chainlit#610) * langchain callback should consider current step * llama index callback should consider current step * add loader while image is loading (Chainlit#612) * enhance custom auth (Chainlit#613) * return error 500 instead of 4xx when feedback fails * remove legacy prod url (Chainlit#615) * Update callbacks.py (Chainlit#614) * fix react client lodash issue (Chainlit#619) * added AWS Cognito OAuth provider (Chainlit#540) (Chainlit#617) * track user sessions (Chainlit#620) * track user sessions * interactive -> is_interactive * update sdk * Adding streaming functionality to output for haystack callback manager (Chainlit#621) * Adding streaming functionality to haystack callback manager * Improved code and added type to self.last_tokens * Fixed another latent bug. Seems like the code would fail when agent_step.is_last() wasn't true as the stack had already been popped * Changed list to List for mypy test (guessing this is due to python version) * fixed mypy issue --------- Co-authored-by: Kevin Longe <kevinlonge3@gmail.com> * do not throw in ws connect handler * Add Chat Settings to Generic Langchain Provider (Chainlit#622) * Add Chat Settings to Generic Langchain Provider * Update langchain.py * Set default inputs to empty list Suggestion from @willydouhard * Do not require settings for stream event * use chainlit 1.0.0 attrs * remove unecessary whitespace * Update langchain.py * Update backend/chainlit/playground/providers/langchain.py Co-authored-by: Willy Douhard <willy.douhard@gmail.com> --------- Co-authored-by: Willy Douhard <willy.douhard@gmail.com> * refactor: Add new way to open thread history (Chainlit#629) * avoid creating existing session (Chainlit#637) * fix element creation (Chainlit#636) * fix element creation * remove print * feat: Add button to auto scroll down (Chainlit#630) * disable jwt auth for elements (Chainlit#633) * feat: New way to stop the loading task (Chainlit#631) * refactor: Design of messages (Chainlit#635) * refactor: Design of messages * ui fixes * fix tests * clear warnings * fix width issue * fix sidebar trigger button * fix vertical centering * enhance user session count * add tooltip to sidebar trigger --------- Co-authored-by: Willy Douhard <willy.douhard@gmail.com> * do not display scroll to bottom button on chat history * fix hide cot hide pp (Chainlit#638) * fix scroll bottom button flickering (Chainlit#639) * literalai (Chainlit#642) * do not display hide of cot toggle is hide cot is true on the server (Chainlit#656) * add onlogout hook (Chainlit#654) * add onlogout hook * pass the fastapi response as well * update the user if it already exists (Chainlit#653) * do not apply a bgcolor on an avatar that has an image (Chainlit#652) * always use the latest uploadFile function (Chainlit#651) * Update server.py (Chainlit#648) * prevent the app to crash if the data layer is not reachable (Chainlit#644) * prevent the app to crash if the data layer is not reachable * make the app still usable if auth is enabled and data layer is down * prepare release (Chainlit#657) * prepare release * bump literalai version * fix parent id issue (Chainlit#659) * prevent running button to display twice when cot is false and streaming (Chainlit#666) * bump version * Added Internationalization with react-i18next (Chainlit#668) * feature: i18n * do not display running button if cot is true and message is being streamed (Chainlit#669) * chore: bump uvicorn to 0.25.0 (Chainlit#664) fixes Chainlit#663 Co-authored-by: = <pawel@Pawels-MacBook-Pro.local> * fix pp doc link * enhance default translation log * attempt to fix file watcher (Chainlit#674) * enhance langchain tracing * fix: Tasklist flick (Chainlit#676) * Wd/embed (Chainlit#679) * relax fast api * allow for custom fonts * make cors configurable * rename to copilot and serve copilot index.js --------- Co-authored-by: SuperTurk <alim.tunc@outlook.fr> * fix translation key * Release/1.0.200 (Chainlit#681) * fix translation issue * fix copilot auth * add continuing chat info * bump version * prepare release * Update README.md * Remove `+` from secrets (Chainlit#688) Some oauth providers will replace the `+` character with a space which fails the auth check. * fix/fix overlay (Chainlit#717) Co-authored-by: Clément Sirieix <clementsirieix@Clements-MacBook-Pro.local> * steps should be able to be called in parallel (Chainlit#694) * steps should be able to be called in parallel * clean local steps * allow generation on message * fix tests * remove fast api version constraint (Chainlit#732) * remove fast api version constraint * bump langchain * update literalai dependency (Chainlit#735) * thread name, wrap step input/output, remove user session * update literal-ai dependency * set the default prompt playground state to `Formatted` * keep the input/output step properties as dict in Chainlit - also fixes the llama index callback and the playground, following the renaming * datalayer: only add input/output if needed, remove empty input/output objects * remove debug log --------- Co-authored-by: Willy Douhard <willy.douhard@gmail.com> * bump literal version (Chainlit#749) * bump literal version * fix pp * Adds Cognito label and icon (Chainlit#704) * adds custom_js (Chainlit#708) * prepare release (Chainlit#751) * fix thread name * fix thread name * fix test * fix mypy * fix resume chat * Add OpenAI integration (Chainlit#778) * Add OpenAI integration - This simplifies instrumenting OpenAI calls - Reuses the literalai OpenAI instrumentation - Creates a new step for each OpenAI call, with the call details in the generation property * add a 1ms delay to the `on_message` callback - This makes sure any children step starts after the parent message step - replaced `generation.settings.model` by `generation.model` * move the openai version check inside the instrumentation call - avoids erroring if a user isn't using `instrument_openai` and hasn't installed openai * Add issue templates (Chainlit#786) - Helps gather more information when users create issues * prepare release (Chainlit#794) * prepare release * fix mypy * fix pdf mime type * Update pyproject.toml to fix python-multipart vulnerability (Chainlit#777) fixes Chainlit#776 * Fix: LlamaIndex 0.10.x migration (Chainlit#797) * fix unbound reference issue (Chainlit#807) * fix unbound reference issue * bump version * Add custom frontend build (Chainlit#783) - Use the `custom_build` to set the relative path for the custom build. - The `custom_build` path should contain a `./frontend/dist/index.html` file that will be load when users load the chainlit app url. - This `index.html` can contain any custom frontend, you're responsible for providing the full ui. * fix: fix typing for data layer (Chainlit#802) * added missing translation for time grouped categories (Chainlit#773) * remove forced lineheight (Chainlit#814) * Multi-modal support for file size, mime types and number of files (Chainlit#787) - adds `accept` (mime type), `max_files` and `max_size_mb` configurations (under `features.multi_modal`) - file uploads are rejected if they aren't respecting the configuration * move react-components codebase back into frontend (Chainlit#829) - simplifies the codebase by removing the need for a separate package * Loads markdown file based on language (Chainlit#692) * Fix 'completion' referenced before assignment (Chainlit#800) - Fixes `UnboundLocalError("local variable 'completion' referenced before assignment")` error in LangchainTracer.on_llm_end callback * fix typo in provider name (Chainlit#822) - `Coginto` -> `Cognito` * add missing translation keys (Chainlit#818) - add the "settings" modal title and the "show history" text * bump react-client version * feat: add HEAD route for '/' to support status checks (Chainlit#835) * Do not reload installed packages on file change (Chainlit#842) - fixes the filewatcher issues - the fix checks where the modules are installed (venv or regular installation) and prevents their reload. * fix BaseDataLayer create_element type (Chainlit#850) * migrate to literal score (Chainlit#851) * migrate to literal score * fix tests * enhance langchain llm step display * correctly display new lines * move new line cleaning to backend * fix thread dict * bump sdk version * fix data layer test * fix casing * allow for custom socketio pathname (Chainlit#853) * Improve translations (Chainlit#852) * add translation linter - compares the local translation json structure with the en-US chainlit reference - helps maintain accurate translations across languages * add translation fallback - when a translation is defined as a country code (ex: `de`), it should be used if the user uses a specific language (ex: `de-DE`) * move translations from settings to its own route - this enables having translations on the login screen (we didn't load the config before a user is logged-in) * add fastapi gzip middleware to speed up json replies * add missing translations - enable a fallback (skeleto component) in the existing Translator component - add a new hook to return a string (with a `...` fallback) - switch to using the Translator as much as possible (to benefit from the skeleton fallback) - if the Translator can't be used, use the new `useTranslation` hook instead of the one from `react-i18next` (to benefit from the fallback which avoids displaying the translation keys) * fix tests * update the copilot translation fetching logic - use the new /project/translations route instead to retrieve the translations * fix the tests * update changelog * Update CHANGELOG.md * bump version * unpin starlette dependency (Chainlit#868) * remove translations that aren't up to date (Chainlit#866) - we won't host all translations in Chainlit as this would lower the speed of iteration (we would need to update each translation for each frontend change that introduces a new translation) - we will need to find a good structure to allow the community to share their translations * Fix dates in changelog (Chainlit#872) * fix: clean sidebar when starting a new conversation (Chainlit#878) - moved the 'SideView' state into react-client - clean up the 'SideView' state in the `clear` method from `useChatInteract` * add optional `tags` and `metadata` to steps and messages (Chainlit#877) * fix tool calls in prompt playground (Chainlit#865) - also get rid of non-chat openai * rename the Literal env variable to LITERAL_API_URL (Chainlit#870) - backward compatible as `LITERAL_SERVER` is used as fallback * add model and chunks to llama index callback handler (Chainlit#876) * Wd/update literalsdk (Chainlit#885) * update to async literal sdk * changelog * make actions trigger thread persistence * add chat profile as thread tag * expose sessionId in useChatSession * Support SQLAlchemy for custom data layer (Chainlit#836) - adds custom, direct database, data layer using SQLAlchemy with support for a wide-range of SQL dialects - configures ADLS or S3 as the blob storage provider - duplicated `PageInfo` and `PaginatedResponse` from literal SDK into backend/chainlit/types.py and updated typing * bump literalai version and chainlit * add quotes to create-secret output (Chainlit#909) - fixes an issue with special characters from the secret * update changelog (Chainlit#910) * bump version * Wd/resilience (Chainlit#913) * make sure the chainlit app works even if the data layer is down * changelog * check that user exists before listing threads * fix fallback * bump literalai version * put back user creation * fix typo * remove return * Add the user's browser languages in the user session (Chainlit#889) * chore: label new issues as "needs-triage" (Chainlit#914) - prevents all untriaged issues from going on the community board * make auto tag thread opt in (Chainlit#927) * fix: correctly serialize generation and allow `None` `storage_provider` in SQLAlchemyDataLayer (Chainlit#921) - adding `self.storage_provider = None` in `def __init__()` so that it does not break method create_element. - adding generation parameter in create_step as it's also a JSON type --------- Co-authored-by: Thibaut Patel <thibaut.patel@gmail.com> * fix flaky parent for openai instrumentation (Chainlit#931) * Allow html in text elements (Chainlit#893) * allow for setting a ChatProfile default (Chainlit#930) * bump literalai sdk version * Update thread list on first interaction (Chainlit#923) - re-fetch the thread list on first interaction - navigate to `/thread/:id` url when creating a new conversation - update the `/thread/:id` page to allow for displaying the current chat - add `threadId` to `useChatMessages` to get the current conversation thread id - update "back to conversation" links - clear current conversation when deleting the current thread or changing chat profile * bump version (Chainlit#932) * fix: pasting from microsoft products generates text instead of an image (Chainlit#934) fixes Chainlit#743 - only consider paste input if it can't be converted as text * add support for `multiline` option in `TextInput` chat settings field (Chainlit#945) closes Chainlit#507 * fix: do not prevent thread revalidation (Chainlit#944) fixes Chainlit#941 - This fixes the issue where only the first messages of a thread would be shown in the thread history in some cases. * fix: display the label instead of the value for menu item (Chainlit#943) * fix: disable gzip middleware to prevent a compression issue on safari (Chainlit#952) fixes Chainlit#895 * release 1.0.506 (Chainlit#953) - update literalai dependency to version 0.0.509 - bump version - update changelog * Wd/audio (Chainlit#962) * rework navigation * wip * fix buffering * finalizing audio feature * fix lint * update changelog * bump literalai version * Run on_chat_resume() before resume_thread() (Chainlit#968) Execute the on_chat_resume decorator before the resume_thread emitter. This allows the `thread` in on_chat_resume to be updated on load. * fix: double on_chat_end invocation (Chainlit#971) * feat: remove bytes objects from steps (Chainlit#969) * feat: remove bytes objects from steps * feat: process list and tuples * Feat/discord (Chainlit#986) * add slack platform * add discord platform * Update sql_alchemy.py (Chainlit#981) * Release/1.1.0 (Chainlit#989) * bump version * fix dependabot security alert * feat: wrap on_message with typing (Chainlit#991) Bot will show as typing while the run is running. * fix user menu overflow * make discord and slack bot resilient to the data layer * prepare 1.1.101 release * fix scroll issues (Chainlit#1000) * Added threadId to Feedback (Chainlit#999) * feat: add custom meta image config (Chainlit#1007) * Wd/1.1.200 (Chainlit#1008) * loader rework * update icons * fix auto scroll * fix github button * update readme * fix details button + loader * fix ci * fix audio capture * fix: linter * feat: add video player using react-player to support YouTube, Vimeo and other sources (Chainlit#980) * Update CHANGELOG.md * feat: add font colors to config.toml (Chainlit#976) - you can now configure the primary and secondary text color from the `config.toml` file. * fix: limit discord thread name (Chainlit#1013) * fix: limit discord thread name * fix: handle start of thread with blank message * update readme * Update README.md * Added threadId to Element (Chainlit#1005) * Added threadId to Feedback * :Revert "Added threadId to Feedback" This reverts commit dd5e4b5. * Add threadId to delete_element * fix: custom build dir do not need frontend and dist (Chainlit#1020) * fix: custom build dir do not need frontend and dist Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * fix e2e test Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * test previous values Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * fix tests Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * no format index.html Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * remove envrc Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> --------- Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> * Release/1.1.300 (Chainlit#1028) * Add starters * Debug mode * Rework CoT * Rework Avatars * Remove PP * Update README.md * update readme image * fix: scroll flickering * 1.1.300rc1 * attempt to fix build (Chainlit#1033) * attempt to fix build * update numpy * condition numpy version based on python version * fix condition on numpy * remove duplicate new chat button * update changelog * feat: added Gitlab OAuth provider (Chainlit#1012) * feat: add DynamoDB Datalayer (Chainlit#1030) - enables using AWS DynamoDB as a database for the chat history * Release/1.1.300 (Chainlit#1040) * enhance conversation spacing and make copilot expandable * fix non ascii characters for chat profiles * Release/1.1.300 (Chainlit#1041) * enhance conversation spacing and make copilot expandable * fix non ascii characters for chat profiles * add input streaming support * Release/1.1.300 (Chainlit#1043) * enhance conversation spacing and make copilot expandable * fix non ascii characters for chat profiles * add input streaming support * fix message margin top * feat: add ssl support using custom key/cert files (Chainlit#1047) --------- Co-authored-by: BCM <dahifi@gmail.com> * Add Teams integration (Chainlit#1003) * add teams integration * enhance feedback and add typing activity * feat: add OAuth Azure AD hybrid flow (Chainlit#1046) * Added Hybrid Flow for Authorization grant to reterive user id_token. Changed Redirect from Get to Post as Hybrid Flow needs form_post instead of query and returns the response as Form body. Added Hybrid Flow for Authorization grant to reterive user id_token. Changed Redirect from Get to Post as Hybrid Flow needs form_post instead of query and returns the response as Form body. * Added suggested changes Added Get for Callback Added Id_token to Callback signature * fixes - add missing random_secret - fix scopes - split the oauth redirection into two routes - fix imports * split azure hybrid flow into a separate oauth provider * fix typing and incorrect merge conflict resolution --------- Co-authored-by: Shabir Jan <sja@matas.dk> * make select input use theme colors * add tooltip to avatar * Update en-US.json (Chainlit#1061) fix small typo * Update sql_alchemy.py - bugfix for delete_step (Chainlit#1027) Fixed bug in sql_alchemy.py:delete_step: Wrong/non-existing primary key name for steps_query (was "forId" instead of id) * prepare release (Chainlit#1064) * Added the ability to specify a root path (sub directory * @cl.on_system_message * add system message * make sure metadata is a dict * add useAudio * fix index error * rc5 --------- Co-authored-by: JT <jtvinolus@gmail.com> * Parametrize uvicorn ws protocol * prepare release (Chainlit#1071) * prepare release * fix: slack bot if user email is not available * fix socketio double cross origin header * add IS_SUBMOUNT * fix: add back get_user_info to AzureADOAuthProvider (Chainlit#1075) * fix: add back get_user_info to AzureADOAuthProvider * prepare patch * prepare release (Chainlit#1081) * prepare release * fix: copilot theme * fix: oauth redirection url (Chainlit#1088) * fix: oauth redirection with root path * feat: add mistral instrumentation (Chainlit#1100) * feat: add mistral instrumentation * fix: remove requirements on mistralai * fix: add changelog * add chat context (Chainlit#1108) * add chat context * Wd/chat context (Chainlit#1109) * add edit_message * fix: clear steps when editing a message * fix: test * prepare release * fix tool nesting (Chainlit#1113) * fix tool nesting * fix: flaky test * Fix Azure authentication (Chainlit#1117) The callback path for authentication in Azure AD, and possibly others, was unintentionally broken in 8ecd415 by modifying the user-facing path. This commit reverts one line of that change to restore it to its previous state. * Willy/cot (Chainlit#1128) * give more options for cot (hidden, tool_call, full) * handle scorable runs at the framework level * lack/Teams/Discord DM threads are now split by day * prepare 1.1.400rc0 * fix: make only last message of a run scorable * fix: data_layer test * enhance run scoring * bump version --------- Signed-off-by: San Nguyen <vinhsannguyen91@gmail.com> Co-authored-by: Willy Douhard <willy.douhard@gmail.com> Co-authored-by: Florian Valeye <florian.valeye@gmail.com> Co-authored-by: DanConstantini <constantinidan@gmail.com> Co-authored-by: Robin Opdam <48250562+ROpdam@users.noreply.github.com> Co-authored-by: datapay-ai <132215461+datapay-ai@users.noreply.github.com> Co-authored-by: Kevin Longe <kevinlonge3@gmail.com> Co-authored-by: Tyler Titsworth <titswortht@gmail.com> Co-authored-by: SuperTurk <alim@refty.co> Co-authored-by: Josh Hayes <35790761+hayescode@users.noreply.github.com> Co-authored-by: Davi Reis Vieira <36394034+DaviReisVieira@users.noreply.github.com> Co-authored-by: Pawel <p.morawian@gmail.com> Co-authored-by: = <pawel@Pawels-MacBook-Pro.local> Co-authored-by: SuperTurk <alim.tunc@outlook.fr> Co-authored-by: Brian Antonelli <brian.antonelli@gmail.com> Co-authored-by: Clément Sirieix <clem.sirieix@gmail.com> Co-authored-by: Clément Sirieix <clementsirieix@Clements-MacBook-Pro.local> Co-authored-by: Thibaut Patel <thibaut.patel@gmail.com> Co-authored-by: Brian Antonelli <brian.antonelli@coxautoinc.com> Co-authored-by: Kevin <kriggen@gmail.com> Co-authored-by: Anurag Dandamudi <dvanurag1308@gmail.com> Co-authored-by: Felipe Aros <felipe.aros.r@gmail.com> Co-authored-by: 131 <11222509+Netrvin@users.noreply.github.com> Co-authored-by: San Nguyen <vinhsannguyen91@gmail.com> Co-authored-by: hans-sarpei <domi_sch94@web.de> Co-authored-by: jpolvto <50706889+jpolvto@users.noreply.github.com> Co-authored-by: steflommen <github@steflommen.com> Co-authored-by: shishax <federico.giuba@gmail.com> Co-authored-by: giulioottantotto <33235839+giulioottantotto@users.noreply.github.com> Co-authored-by: Sinan Saral <sinansaral@gmail.com> Co-authored-by: BCM <dahifi@gmail.com> Co-authored-by: rickythefox <richard@ginzburg.se> Co-authored-by: Koichi Ishida <wapa5pow@gmail.com> Co-authored-by: mohamedalani <mohamed.alani76@gmail.com> Co-authored-by: Jan Beitner <jan.beitner@gmail.com> Co-authored-by: Kevin Merritt <kevinmerritt@outlook.com> Co-authored-by: Kevin Merritt <kevinwmerritt@gmail.com> Co-authored-by: Maciej Wieczorek <maciej.wieczorek@gmail.com> Co-authored-by: Mayaank Vadlamani <mayaankvad@gmail.com> Co-authored-by: Quy Tang <3761730+qtangs@users.noreply.github.com> Co-authored-by: Hugues de Saxcé <hugues@chainlit.io> Co-authored-by: DanConstantini <16107237+constantinidan@users.noreply.github.com> Co-authored-by: Mathieu CHANIAT <matchaniat@aol.com> Co-authored-by: Shabir Jan <sja@matas.dk> Co-authored-by: Davi S. Zucon <davizucon@users.noreply.github.com> Co-authored-by: ralphkink <rpkink@hotmail.com> Co-authored-by: JT <jtvinolus@gmail.com> Co-authored-by: Maciej Wieczorek <mwieczorek@evertz.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
fixes #663