投稿時間:2022-05-03 07:22:35 RSSフィード2022-05-03 07:00 分まとめ(41件)

カテゴリー等	サイト名等	記事タイトル・トレンドワード等	リンクURL	頻出ワード・要約等/検索ボリューム	登録日
AWS	AWS	Why am I being billed for Elastic IP addresses when all my Amazon EC2 instances are terminated?	https://www.youtube.com/watch?v=FHyo75YjrcE	Why am I being billed for Elastic IP addresses when all my Amazon EC instances are terminated Skip directly to the demo For more details see the Knowledge Center article with this video Disha shows you why you are being billed for Elastic IP addresses when all your Amazon EC instances are terminated Introduction Demo ClosingSubscribe More AWS videos More AWS events videos ABOUT AWSAmazon Web Services AWS is the world s most comprehensive and broadly adopted cloud platform offering over fully featured services from data centers globally Millions of customers ーincluding the fastest growing startups largest enterprises and leading government agencies ーare using AWS to lower costs become more agile and innovate faster AWS AmazonWebServices CloudComputing	2022-05-02 21:23:33
AWS	AWSタグが付けられた新着投稿 - Qiita	脳筋が岡山県内のジムを検索するサービスを開発しました(ポートフォリオ、Rails+AWS+Docker)	https://qiita.com/omeinu/items/b1a22c14c6b35b45e287	gymseek	2022-05-03 06:44:20
Docker	dockerタグが付けられた新着投稿 - Qiita	脳筋が岡山県内のジムを検索するサービスを開発しました(ポートフォリオ、Rails+AWS+Docker)	https://qiita.com/omeinu/items/b1a22c14c6b35b45e287	gymseek	2022-05-03 06:44:20
Ruby	Railsタグが付けられた新着投稿 - Qiita	脳筋が岡山県内のジムを検索するサービスを開発しました(ポートフォリオ、Rails+AWS+Docker)	https://qiita.com/omeinu/items/b1a22c14c6b35b45e287	gymseek	2022-05-03 06:44:20
海外TECH	Ars Technica	Apple clarifies its controversial app removal emails with a clearer policy statement	https://arstechnica.com/?p=1851724	notice	2022-05-02 21:29:19
海外TECH	DEV Community	Compose Hackathon: Day 1	https://dev.to/zachklipp/compose-hackathon-day-1-2ja3	Compose Hackathon Day IntroductionThis week the Jetpack Compose team is doing a hackweek I am using it as an excuse to try building something I ve been thinking about basically since I joined the team I m not sure if it will work or if it will be fast enough to do what it needs to do but I m excited about the possibilities it opens up if it turns out To document my plan and my findings I ll use this blog series as a journal hopefully posting once a day but we ll see how things go Since I want to focus on the actual work please excuse the lack of editing and polish MotivationCompose s text editing APIs currently have the following shape for communicating the text being edited and changes to that text TextField text String onTextChanged String gt Unit However internally there are many sophisticated operations that need to be performed on the current text to make changes Unlike simple hardware keyboards software keyboards like Android s IME are aware of the contents of the text fields and can perform operations on more than one character at a time eg when accepting an auto correct suggestion These commands can be delivered in batches and so internally compose turns the text value into a mutable buffer that is more efficient to apply operations to When a batch of IME commands comes in it applies them all to the internal buffer at once then converts that buffer to a string to pass back to the callback It keeps this buffer around between compositions to optimize for the case where the callback immediately accepts the text and provides the exact new text back in the next composition Then on every composition it has to make sure the received text matches what it thinks it should be otherwise it has to reset its internal state to match The reason this simple API was chosen is because it is simple to teach understand and work with while still allowing state hoisting andimmutable values it doesn t introduce any new types and it matches many other APIs in compose However if a developer wants to talk directly to the low level IME API they have to create their own buffer type Kotlin doesn t provide any efficient text editing buffers which need to support efficient sequential inserts and deletes in the middle of the string And even if a developer simply wants to observe text changes or perform their own transformations there s no way for example to actually determine what text was inserted deleted without comparing the before after strings which is ridiculous because the internal code already knows this so it shouldn t be necessary to recompute it It s become apparent that this simple “value change callback pattern is almost too simple While there has been a big paradigm shift in the android world over the last years to preferring immutable data for many good reasons I think a solution to these problems might be to embrace mutability Compose s snapshot state system provides tools that address many of the reasons why we ve all become afraid of mutability and I think if we lean into that it might produce APIs that are still easy to use provide decoupled API boundaries and might even have better performance The projectWhat if instead of passing a value change callback around we just passed a mutable text buffer This buffer would be defined as an interface that supports a minimal set of mutation and read operations We would provide a performant default implementation similar to how Kotlin provides an ArrayList as the underlying data structure when you ask for a mutableListOf But developers could create their own wrappers if they want to have more control This isn t currently very practical because Compose uses a private gap buffer for its text editing This is an efficient data structure for such a use case but it requires careful handling because it doesn t participate in Compose s state snapshots It s basically an ArrayList for text editing you wouldn t use an ArrayList in regular Compose code you d use a snapshot aware implementation like the one created by mutableStateListOf So if we want to make a mutable text buffer part of our public API we need to make a buffer that knows about snapshot state so it s safe to use in regular compose code See Goals below for an enumeration of the advantages of a snapshot aware buffer As mentioned above Compose currently uses a gap buffer But there are two other well known data structures that are used for this purpose the rope and the piece table When the Compose team was writing the first text field code one of my team members wrote a nice brief introduction and evaluation of all three which is unfortunately an internal document so I can t share it but there is a lot of public literature about them see the Related Reading section at the bottom of this post But of all three I think the piece table is the most well suited to adapting to be aware of snapshots And if we can do that then we can start passing around mutable buffers in text editing and IME APIs reusing the intuitions that Compose developers have built up around how state in Compose is managed and how changes are communicated GoalsBuild a piece table that is unit tested and supports basic editing and read operations We can probably make it implement standard platform agnostic interfaces like CharSequence for reads and Appendable StringBuilder for writes Write or adapt benchmarks to compare this implementation to the current gap buffer as well as to allow profiling and optimizing our implementation Adapt the piece table to have the following snapshot related features these should sound familiar as most of them are how other snapshot state classes work Mutations to a buffer should not be visible outside the current snapshot until the snapshot is committed Mutations to a buffer should be discarded if the snapshot is discarded before committing Snapshot observers that read a buffer should be notified when the buffer is changed that is when the snapshot that changes it is committed Multiple threads should be able to mutate the same buffer independently in their own snapshots without explicit synchronization other than snapshots A sequence of mutations performed in the same snapshot should not have very much overhead Stretch GoalsIf time permits Optimize the snapshot aware piece table s performance as much as possible using the benchmarks Explore how to use this data structure to build something like AnnotatedString Builder for editing text that has spans associated with it Explore how to implement undo redo operations using the piece table Non goalsSince this hackathon is only a week long I want to use it to prove that we can build such a data structure and keep the scope narrow If it looks promising future work can be prioritized and performed outside of the hackathon we already have bugs filed for many of those tasks Change text field or IME APIs or implementation to use this data structure in production code Productionize the piece table itself in any way The data structureSince this post is already getting a bit long I m not going to explain piece tables in depth see the Related Reading section at the bottom of this post for some excellent explainers but just as a quick refresher a piece table consists of Two text values an immutable one that holds the initial value of the text when the table was created and an append only one that stores all new text inserted or added into the buffer A table of source index length tuples that describe which substrings to take from each of the two text values in in order to construct the current buffer value To make this data structure snapshot aware we need to make both parts of the data structure snapshot aware One of the nice things about the snapshot state system is that any data structure composed of snapshot safe parts is itself snapshot safe The second part should be relatively straightforward we just need to store the table in a SnapshotStateList The first part might be more challenging How do we make an array snapshot friendly The entire array needs to be efficiently accessible by index so we can t just store a SnapshotStateList of each newly appended string We also want to make sure that a sequence of operations performed in the same snapshot can be performed without much overhead neither computational nor space Here s how I see this working I don t know if it will actually work so the plan might change as I actually build it We can break the single array into a list of fixed length blocks similar to Okio s Segments Every block except the last one will be full Then we can use a SnapshotStateList to hold the blocks This lets us copy the whole array efficiently because the underlying implementation of SnapshotStateList is a persistent data structure so most blocks can be shared between copies To make a copy for mutation we only need to copy the actual contents of the last block which will only be partially filled since that s the only block that will actually be mutated it s append only We ll also store some metadata that lets us determine if any snapshots other than the current one have access to the current list of blocks When performing a write operation we ll check if any other snapshots can see our last block If not then the current snapshot has exclusive access to the block so we can just write directly into the block s array The size of the contents of the last block must also be stored in a snapshot state object so that if the snapshot is discarded after a mutation the new size is also discarded note that the discarded data will still exist in the block but it will be overwritten on the next mutation However if another snapshot can see the last block we can t write to it because another snapshot may also write into it so we need to copy the last block s contents into a new array and replace the last element in the list of blocks Since the block list is a SnapshotStateList we can just use an indexed replace operation for the last block to do this efficiently Then once we have our own dedicated block subsequent mutations will hit the first case above and just write directly into the array If the snapshot is discarded the copy we just made of the block will be discarded and eventually garbage collected The only part of this design that I m not confident about is being able to determine if the current snapshot has exclusive access to the last block I need to go refresh myself on the particulars of MVCC but I think it might be possible by storing the ID of the snapshot that created the block and comparing it to the current snapshot ID This might require dropping down into the low level APIs for actually creating and managing snapshot state records which both scares me and makes me excited for a new learning exercise Because the size of the last block is snapshot aware multiple reading snapshots can actually share access to it We only need to ensure that a single snapshot has write access the readers wouldn t see the size change until the mutating snapshot is committed I m not sure if it s possible to make that distinction Stay tuned…That s it for day one Keep checking back to follow my progress I ll be adding more posts to this series over the week Related ReadingDarren Burns piece table tutorialText Buffer Reimplementation a Visual Studio Code Story	2022-05-02 21:14:26
海外TECH	DEV Community	End-to-End Monitoring with Grafana Cloud with Minimal Effort	https://dev.to/martinheinz/end-to-end-monitoring-with-grafana-cloud-with-minimal-effort-13a1	End to End Monitoring with Grafana Cloud with Minimal EffortMonitoring is usually at the end of a checklist when building an application yet it s crucial for making sure that it s running smoothly and that any problems are found and resolved quickly Building complete monitoring including aggregated logging metrics tracing alerting or synthetic probes requires a lot of effort and time though not to mention building and managing the infrastructure needed for it So in this article we will look at how we can set all of this up quickly and with little effort with no infrastructure needed using Grafana Cloud all for free Infrastructure ProviderSmall or medium sized projects don t warrant spinning up complete monitoring infrastructure so managed solutions are a good choice At the same time no one wants to spend a lot of money just to keep a few microservices alive There are however a couple managed monitoring solutions with free tiers that provide all the things one might need First of them being Datadog which provides free infrastructure monitoring which wouldn t get us very far especially considering that alerts are not includedBetter option is New Relic which has a free plan which includes probably everything that you might need One downside is that New Relic uses a set of proprietary tools which creates a vendor lock and would make it hard to migrate to another platform or to own infrastructure Third and in my opinion the best option here is Grafana Cloud which has a quite generous free plan that includes logging metrics alerting and synthetic monitoring using a popular set of open source tools such as Prometheus Alertmanager Loki and Tempo This is the best free platform I was able to find and it s what we will use to set up our monitoring For a sake of completeness I also looked Dynatrace Instana Splunk Sumo Logic and AWS Managed Prometheus those however have no free plans As I mentioned I m considering only free options If you need to run a monitoring stack on your infrastructure I strongly recommend to use Prometheus and friends that is Prometheus Alertmanager Thanos and Grafana The easiest deployment option for that is using Prometheus Operator Note This isn t sponsored heh I wish I just decided to take the platform for a spin and I really liked it so I decided to make write up for you and my future self MetricsYou can sign up for Grafana Cloud account at assuming you ve done so you should now have your account accessible at We will start building our monitoring by sending Prometheus metrics to the account Before we begin though we need to get access to the Prometheus instance provisioned for us by Grafana Cloud You can find instances of all the available services in your account on the Grafana Cloud Portal at From there you can navigate to Prometheus configuration by clicking Details button There you will find all the info needed to send data to your instance that is username remote write endpoint and API Key which you need to generate Usually Prometheus scrapes metrics in Grafana Cloud the Prometheus instance is configured to use push model where your application has to push metrics using Grafana Cloud Agent On the above mentioned Prometheus configuration page you re also presented with sample remote write configuration which you can add to your agent To this out we will spin up a simple application and agent using docker compose docker compose ymlversion services agent image grafana agent v container name agent entrypoint bin agent config file etc agent agent yaml metrics wal directory tmp wal config expand env config enable read api environment HOSTNAME agent PROMETHEUS HOST PROMETHEUS HOST PROMETHEUS USERNAME PROMETHEUS USERNAME PROMETHEUS PASSWORD PROMETHEUS PASSWORD volumes PWD agent data etc agent data PWD agent config agent yaml etc agent agent yaml ports api image quay io brancz prometheus example app v container name api ports expose The above config provides both the agent and sample Go application with reasonable default settings It sets the Prometheus host username and API Key password through environment variables which should be provided using env file Along with the above docker compose yml you will also need agent configuration such as agent config agent yamlserver log level info http listen port metrics wal directory tmp wal global scrape interval s configs name api scrape configs job name default metrics path metrics static configs targets api remote write basic auth username PROMETHEUS USERNAME password PROMETHEUS PASSWORD url https PROMETHEUS HOST api prom pushThis config tells the agent to scrape the sample application running at api with metrics exposed at metrics It also tells the agent how to authenticate to Prometheus when pushing metrics In a real world application you might be inclined to run the metrics endpoint over HTTPS that however won t work here so make sure that the server listens on HTTP not HTTPS Finally after running docker compose up you should be able to access the API metrics with curl localhost metrics and also see logs of agent such as agent ts T Z caller node go level info agent prometheus component cluster msg applying config agent ts T Z caller remote go level info agent prometheus component cluster msg not watching the KV none set agent ts T Z level info caller traces traces go msg Traces Logger Initialized component tracesagent ts T Z caller server go level info msg server configuration changed restarting server agent ts T Z caller gokit go level info http grpc msg server listening on addresses agent ts T Z caller wal go level info agent prometheus instance msg replaying WAL this may take a while dir tmp wal walagent ts T Z caller wal go level info agent prometheus instance msg WAL segment loaded segment maxSegment agent ts T Z caller dedupe go agent prometheus instance component remote level info remote name url msg Starting WAL watcher queue agent ts T Z caller dedupe go agent prometheus instance component remote level info remote name url msg Starting scraped metadata watcher agent ts T Z caller dedupe go agent prometheus instance component remote level info remote name url msg Replaying WAL queue agent ts T Z caller dedupe go agent prometheus instance component remote level info remote name url msg Done replaying WAL duration sTo also view what metrics the agent is itself collecting and sending you can use curl localhost metrics DashboardsWith the data flowing to your Prometheus instance it s time to visualize it on a dashboards Navigate to and click New Dashboard and New Panel in the following screen Choose grafanacloud lt USERNAME gt prom as a data source and you should see metrics browser field getting populated with your metrics Below you can see a sample dashboard showing memory consumption of a Go application The dashboard was additionally configured to show a threshold of MB which can be set in the bottom of right side panel SyntheticsIn addition to monitoring your applications using metrics they expose you can also leverage synthetic monitoring that allows you to continuously probe the service for status code response time DNS resolution etc Grafana Cloud provides synthetic monitoring which can be accessed at From there you can navigate to Checks tab and click Add new check There you can choose from HTTP PING DNS TCP or Traceroute options which are explained in docs Filling out the remaining fields should be pretty self explanatory when choosing probe location though be aware that the more probes you run the more logs will be produced which count towards your consumption usage limit Nice feature of Grafana Cloud Synthetics is that you can export them as Terraform configuration so you can build the checks manually via UI and get the configuration as a code To do so navigate to and click Generate config button After you are done creating your checks you can view dashboards that are automatically created for each type of check For example HTTP check dashboard would look like this If you re monitoring website that has web analytics configured then you will want to exclude IPs of Grafana probes from analytics collection There s unfortunately no list of static IPs however you can use the following DNS names to find the IPs Alerts and NotificationsBoth metrics and synthetic probes provide us with plenty of data which we can use to create alerts To create new alerts you can navigate to and click New alert rule If you followed the above example with Prometheus then you should choose grafanacloud lt USERNAME gt prom as data source You should already see query fields prepared you can put your metric in the field A and the expression to evaluate the rule in field B The complete configuration should look like so When you scrolled through the available metrics you might have noticed that it now also includes fields such as probe all success sum generally probe these are metrics generated by synthetic monitors shown in previous section and you can create alerts using those too Some useful examples would be SSL Expiration probe ssl earliest cert expiry instance job Ping Website probe Frankfurt time Use Condition WHEN last OF A IS BELOW N Days API Availability sum increase probe all success sum instance some url com job Ping API m OR increase probe success sum instance some url com job Ping API m sum increase probe all success count instance some url com job Ping API m OR increase probe success count instance some url com job Ping API m Use condition WHEN avg OF A IS BELOW XPing Success Rate avg over time probe all success count instance some url com m avg over time probe all success sum instance some url com m Use condition WHEN avg OF A IS BELOW XAgent Watchdog up instance backend Use condition WHEN last OF A IS BELOW You can also use the existing synthetic monitoring dashboards for inspiration when you hover over any panel in the dashboard you will see the PromQL query used to create it You can also go into dashboard settings make it editable and then copy the queries from each panel With alerts ready we need to configure a Contact points to which they will send notifications Those can be viewed at By default there s an email contact point created for you you should however populate its email address otherwise you won t receive the notifications You can add more contact points by clicking New contact point Finally to route the alert to the correct contact point we need to configure Notification policies at otherwise it would all go the default email contact Click New Policy and set the contact point of your choosing optionally also set matching labels if you want to assign only a subset of alerts to this contact In my case I set the label to channel slack both on alerts and the policy And here are the resulting alerts send to email and Slack respectively If you re not a fan of the default template of the alert messages you can add your own in Contact points tab by clicking New template LogsGrafana Cloud also allows you to collect logs using Loki Instance of which is automatically provisioned for you To start sending logs from your applications to the Grafana Cloud you can head to the Grafana Cloud Portal and retrieve credentials for Loki same as with Prometheus earlier You can either use the agent to send the logs or if you re running the apps with Docker then you can use Loki logging plugin which we will do here First you will need to install the plugin docker plugin install grafana loki docker driver latest alias loki grant all permissionsdocker plugin lsID NAME DESCRIPTION ENABLEDd loki latest Loki Logging Driver trueAfter which you need to update the docker compose yml like so version x logging amp default logging driver loki options loki url https LOKI USERNAME LOKI PASSWORD LOKI HOST loki api v push services agent image grafana agent v container name agent logging default logging entrypoint environment LOKI HOST LOKI HOST LOKI USERNAME LOKI USERNAME LOKI PASSWORD LOKI PASSWORD volumes PWD agent data etc agent data PWD agent config agent yaml etc agent agent yaml ports api image quay io brancz prometheus example app v logging default logging container name api ports expose The above sets the logging driver provided by the Docker Loki plugin for each container If you want to set the driver for individual containers started with docker run then checkout docs here After you start your containers with updated config you can verify whether agent found logs by checking tmp positions yaml file inside the agent container With logs flowing to Grafana Cloud you can view them in by choosing grafanacloud USERNAME logs as a data source and querying your project and containers TracesFinal piece of puzzle in the monitoring setup would be traces using Tempo Again similarly to Prometheus and Loki config you can grab credentials and sample agent configuration for Tempo from Grafana Cloud Portal The additional config you will need for the agent should look like All config options at traces configs name default remote write endpoint TEMPO HOST basic auth username TEMPO USERNAME password TEMPO PASSWORD receivers otlp protocols http Additionally docker compose yml and env should include credentials variables for TEMPO HOST TEMPO USERNAME and TEMPO PASSWORD After restarting your containers you should see in agent logs that the tracing component got initialized agent msg Traces Logger Initialized component tracesagent msg shutting down receiver component traces traces config defaultagent msg shutting down processors component traces traces config defaultagent msg shutting down exporters component traces traces config defaultagent msg Exporter was built component traces traces config default kind exporter name otlp agent msg Exporter is starting component traces traces config default kind exporter name otlp agent msg Exporter started component traces traces config default kind exporter name otlp agent msg Pipeline was built component traces traces config default name pipeline name tracesagent msg Pipeline is starting component traces traces config default name pipeline name tracesagent msg Pipeline is started component traces traces config default name pipeline name tracesagent msg Receiver was built component traces traces config default kind receiver name otlp datatype tracesagent msg Receiver is starting component traces traces config default kind receiver name otlpagent msg Starting HTTP server on endpoint component traces traces config default kind receiver name otlpagent msg Setting up a second HTTP listener on legacy endpoint component traces traces config default kind receiver name otlpagent msg Starting HTTP server on endpoint component traces traces config default kind receiver name otlpagent msg Receiver started component traces traces config default kind receiver name otlpIn the above output you can see that the collector is listening for traces at You should therefore configure your application to send the telemetry data to this endpoint See OpenTelemetry reference to find variables relevant for your SDK To verify that traces are being collected and send to Tempo you can run curl s localhost metrics grep traces which will show you following metrics TYPE traces exporter enqueue failed log records countertraces exporter enqueue failed log records exporter otlp traces config default HELP traces exporter enqueue failed metric points Number of metric points failed to be added to the sending queue TYPE traces exporter enqueue failed metric points countertraces exporter enqueue failed metric points exporter otlp traces config default HELP traces exporter enqueue failed spans Number of spans failed to be added to the sending queue TYPE traces exporter enqueue failed spans countertraces exporter enqueue failed spans exporter otlp traces config default HELP traces exporter queue size Current size of the retry queue in batches TYPE traces exporter queue size gaugetraces exporter queue size exporter otlp traces config default Closing ThoughtsInitially I was somewhat annoyed by the day Pro Trial as I wanted to test the limits of the Free plan However after the trial expired I realised that I didn t even exceed the free plan limit and didn t touch the Pro paid features so the Free plan seems to be quite generous especially if you re just trying to monitor a couple of microservices with low ish traffic Even with the free plan Grafana Cloud really provides all the tools you need to set up complete monitoring at actual zero cost for reasonably large deployments I also really like that the usage is calculated only based on amount of metric series and log lines making it very easy to track This is also helped by comprehensive Billing Usage dashboards	2022-05-02 21:01:12
海外TECH	Engadget	Google's latest Pixel 6 and 6 Pro update fixes weak haptic feedback for notifications	https://www.engadget.com/google-pixel-6-may-update-211042700.html?src=rss	Google x s latest Pixel and Pro update fixes weak haptic feedback for notificationsGoogle s recent Pixel software updates haven t always landed flawlessly At the end of last year for instance the company was forced to pause the release of an OTA after reports that the software caused the Pixel and Pro to drop calls More recently the March update introduced an issue that left the company s latest phones producing much weaker notifications when you got a notification Many Pixel and Pro owners complained after Google released the update noting that no matter what they set their phone s haptic feedback to they would miss calls and emails because they couldn t feel their device vibrating Our May software update is now rolling out to supported Pixel devices The update includes Improvements for haptic feedbackFixes for display amp launcherLatest security fixesDevice applicability variesLearn more on our Community post ーMade By Google madebygoogle May On Monday Google began rolling out the May Pixel software update It includes a fix for the vibration issue “Improvements for haptic feedback under certain conditions and uses cases the company writes on its community website The update resolves two other issues that affect all recent Pixels devices The first involves a bug that had caused those phones to wake their displays without any input The second solves a problem that could crash the Pixel launcher after you restarted your device The update also includes the latest Android security patch from Google According to Google it will roll out the May update to all eligible Pixel devices in the coming weeks If you re feeling adventurous you can attempt to install the software on your phone by manually sideloading it Just note that flashing a device always comes with a degree of risk	2022-05-02 21:10:42
海外科学	NYT > Science	How to Watch the Rocket Lab Launch Today	https://www.nytimes.com/2022/05/02/science/rocket-lab-launch-helicopter.html	How to Watch the Rocket Lab Launch TodayIf Rocket Lab can snatch its spent rocket booster from the sky and then reuse it for another orbital launch it will pull off something so far achieved only by Elon Musk s SpaceX	2022-05-02 21:43:10
海外科学	NYT > Science	Which Animal Viruses Could Infect People? Computers Are Racing to Find Out.	https://www.nytimes.com/2022/04/27/science/pandemic-viruses-machine-learning.html	Which Animal Viruses Could Infect People Computers Are Racing to Find Out Machine learning is known for its ability to spot fraudulent credit charges or recognize faces Now researchers are siccing the technology on viruses	2022-05-02 21:32:59
ニュース	BBC News - Home	Russia attacking Mariupol steelworks after evacuations, says Ukraine commander	https://www.bbc.co.uk/news/world-europe-61296851?at_medium=RSS&at_campaign=KARANGA	azovstal	2022-05-02 21:38:21
ニュース	BBC News - Home	World Snooker Championship 2022: Ronnie O'Sullivan claims record-equalling seventh world title	https://www.bbc.co.uk/sport/snooker/61294622?at_medium=RSS&at_campaign=KARANGA	World Snooker Championship Ronnie O x Sullivan claims record equalling seventh world titleRonnie O Sullivan claims his seventh World Championship title with an win over Judd Trump to equal Stephen Hendry s record in the modern era	2022-05-02 21:13:56
ニュース	BBC News - Home	Israel outrage at Sergei Lavrov's claim that Hitler was part Jewish	https://www.bbc.co.uk/news/world-middle-east-61296682?at_medium=RSS&at_campaign=KARANGA	blood	2022-05-02 21:07:12
ニュース	BBC News - Home	Jacky Hunt-Broersma: The cancer survivor who ran 104 marathons in 104 days	https://www.bbc.co.uk/news/world-us-canada-61299527?at_medium=RSS&at_campaign=KARANGA	record	2022-05-02 21:38:57
ニュース	BBC News - Home	Man Utd 3-0 Brentford: Fernandes, Ronaldo & Varane score in morale-boosting win	https://www.bbc.co.uk/sport/football/61212449?at_medium=RSS&at_campaign=KARANGA	Man Utd Brentford Fernandes Ronaldo amp Varane score in morale boosting winManchester United claim a morale boosting victory over Brentford to end a run of three Premier League matches without a win	2022-05-02 21:11:37
ニュース	BBC News - Home	Madrid Open: Andy Murray beats Dominic Thiem to reach round two	https://www.bbc.co.uk/sport/tennis/61301558?at_medium=RSS&at_campaign=KARANGA	madrid	2022-05-02 21:26:52
北海道	北海道新聞	大谷、代打で二ゴロ　鈴木、筒香は試合なし	https://www.hokkaido-np.co.jp/article/676742/	大リーグ	2022-05-03 06:33:00
北海道	北海道新聞	ＥＵ、ルーブル払い拒否を再確認　ロシア産ガスでエネ相会合	https://www.hokkaido-np.co.jp/article/676739/	拒否	2022-05-03 06:02:00
北海道	北海道新聞	事務所離れ船長と連絡せず　社長、安全管理規程に違反	https://www.hokkaido-np.co.jp/article/676741/	離れ	2022-05-03 06:02:00
北海道	北海道新聞	犠牲者の長男「全員見つかって」　観光船事故、代表取材に心境	https://www.hokkaido-np.co.jp/article/676740/	知床半島	2022-05-03 06:02:00
ビジネス	東洋経済オンライン	GWに熟読！東大生推薦｢頭がよくなる参考書｣5選 丸暗記ではなく､そうなる理由もセットで学ぶ | 生まれつきの才能は不要 東大｢逆転合格｣の作法 | 東洋経済オンライン	https://toyokeizai.net/articles/-/586534?utm_source=rss&utm_medium=http&utm_campaign=link_back	東洋経済オンライン	2022-05-03 06:37:00