Announcements（公告）

Pipeline Builder: Next-gen data integration

Date published: 2022-08-18

Designed to enable fast, flexible, and scalable delivery of data pipelines

Building pipelines at scale is often a costly and gated endeavor, requiring extensive technical knowledge, budget, and time allocation to ensure healthy, comprehensive pipelines. With such baseline requirements, data integration can be inaccessible to many users, costly to conduct, and incredibly difficult to maintain.

We created Pipeline Builder so users can focus on describing the pipeline they want to build without worrying about the details.

Building pipelines at speed

Pipeline Builder provides managed, serverless support for batch, micro-batch, and streaming workflows across the full range of target outputs, while builds, refreshes, and other orchestrations are handled together at scale.

• Immediately apply transforms to data (structured, unstructured, IoT) without needing to spin up environments or write boilerplate code.
• Receive instant feedback on failed builds without needing time- and resource-exhaustive checks through entire datasets.
• Debugging is made easier with granular column provenance across an entire pipeline.
• Benefit from Foundry’s world-class security primitives, including purpose-based access, Palantir’s Cipher for selective encryption, and more.
• Pipeline Builder only materializes data necessary to fulfill the specified targets of pipelines, enabling savings on compute and storage at scale.

Streamlined data integration without sacrificing versatility

Builder is designed to fit any data integration need with hundreds of commonly used functions as well as the capability to easily add new ones. Pipeline Builder is also back-end extensible, with the ability to employ different execution engines, (e.g., in-memory engines for small-scale data; Spark for classic batch-based pipelines; Flink for low-latency, streaming pipelines). Builder’s backend ensures a consistent experience regardless of underlying compute modality.

Additionally, multiple users can now provide input through the pipeline build process, independent of considerations on their technical ability. Collaboration and export to code are such essential factors built-in to help ideate, create, and review. If needed, you can use the "Export to code" feature to obtain a Java Spark version of your pipeline.

When should I use Pipeline Builder?

Pipeline Builder creates end-to-end pipelines from data sources to final outputs using Palantir’s comprehensive suite of built-in data transformations. Use it when integrating data towards analysis and application building.

If code is required for highly customized integrations, Authoring can be used to apply transformations on your data using SQL, Python and Java. For most workflows, Pipeline Builder and Authoring complement each other, with the outputs of an Authoring repository being accepted as inputs to Builder pipelines, and vice versa. Both tools offer collaboration through version control. Following integration, for analytics, we’d recommend tools such as:

• Code Workbooks: Use the interactive console for rapid iteration on transformation logic, ad-hoc data exploration, creation of detailed, interactive images with commonly used packages (Matplotlib, Plotly, Seaborn).
• Quiver: Use for search and visualization of native time series analysis and object data. Leverage advanced analytical workflows in the Ontology layer through a visual point-and-click interface and a powerful charting library. Quiver analyses can be templatized into read-only dashboards for broader consumption.

Pipeline Builder is a powerful data integration tool that can encompass the end-to-end creation and management of your pipeline. To learn more about Pipeline Builder, watch our launch video ↗, or review our documentation on Pipeline Builder.

---

Notepad: Modern reporting in Foundry

Date published: 2022-08-15

Quickly share insights and automate report generation

Notepad is a rich text editor that empowers users to share insights generated in the Foundry platform and automate ongoing report generation. Reports generated by Notepad are backed by Foundry’s ontology, allowing users to quickly understand the history and context of any data source used in the document. With historical lineage, your team can readily explain past report versions or decisions by tracing original inputs. Additionally, a Notepad document further helps connect the dots by showing relevant or related content that used the same input data.

With the readily-available context and knowledge, users are empowered to build workflows on previous findings instead of starting entirely from scratch.

Next generation reporting with Notepad

Notepad is a rich text editor with the ability to embed a growing number of sources from across the entire Foundry platform - including Contour, Quiver, Foundry Code Workbook, and Object Explorer. Notepad enables document templatization, allowing you to create the latest version of your periodic reports in just one click.

• Embed charts, tables, and graphs from other Foundry applications. Read about all the supported integrations on our public documentation.
• Maintain structured links to embedded objects and resources which automatically become part of the underlying ontology. Clear lineage around the inputs to any given document supports reproducibility and comprehensibility.
• Freeze content to capture point-in-time context.
• Templatize documents to easily create report versions based on new inputs.
• Export to PDF and print all Notepad documents with granular control over export presentation (embed appearance, page breaks, etc.).

Using Notepad to its fullest potential

Notepad was designed to cover a variety of use cases ranging from ad-hoc to templated periodic reporting:

• Ad-hoc note-taking: Integrate process artifacts and results from other Foundry analytics applications like Quiver. Copy and paste charts and tables to allow for persisting and precise documentation.
• In-platform documentation: Deep integration with platform resources support the direct creation of documentation.
• Periodic reporting: Use templates to create blueprints that allow you to generate an up-to-date report in one click.
• Templated exports: Documents and embedded artifacts are optimized for export. Use templates to generate and download an export-friendly PDF for sharing.

Notepad's robust integration with Foundry apps and flexible editing options help it meet your needs across a variety of use cases.

When should I use Notepad?

Use Notepad to consolidate insights and create documents that are intuitive to build and consume. Consider your end output to discern the Foundry ecosystem tool most suitable for you: Notepad is recommended for the majority of reporting use cases based on live-updating and static documents. To perform dynamic filtering based on user inputs or chart selections, such as a dashboard to present analytical findings, consider:

• Quiver dashboards for object or time-series data
• Contour dashboard mode for tabular data
• Workshop for low-code or no-code and dashboards more advanced operational applications

If you want to learn how to seamlessly report on and build documents that integrate across products in the Foundry ecosystem, see the Notepad documentation.

---

Slate: Global stylesheets

Date published: 2022-08-11

Slate global stylesheets provide application builders a one-stop-shop for styling multiple Slate applications.

Slate enables users to create and apply custom styling to any application, but maintaining consistent styles across a variety of applications in your organization can be challenging and resource intensive. With global stylesheets, app builders can apply their custom branding across all Slate applications with just one click.

One style, everywhere

You can now create a tight, consistent, and recognizable visual brand, even as several people build out applications.

As stakeholders often use more than one Slate application in a workflow to dive into data, builders can now use global stylesheets to craft a consistent style across all applications to ensure a uniform look and feel. Build a global stylesheet whenever you need to introduce application consistency and accessibility across your user base.

• Multiple builders can access design libraries custom-designed for the same organization.
• Select a global stylesheet, and toggle on "Use in application" to have any application in the same namespace adapt the same underlying design guidelines.
• Build app design as part of a one-stop-shop process in a product-supported way.
• Embed accessible UI elements just once, with ease.

Build a global stylesheet for each desired type of view to have mix-and-match applicability

For example, this might mean one global stylesheet for maps, another for inbox views, and one more for charts or graphs. Where two or more of the same components are used, multiple stylesheets can be toggled on for combination use. For example, if you have a Slate dashboard, you might apply the "dashboard" styling. If you have an application better viewed in dark mode, you can select a "dark mode" styling if the application builder has finished it. Similarly, apps for one organization can have the same branding and feel adapted.

To pick between designing a local or global stylesheet, consider your style needs

If you are creating for a workflow that uses multiple applications end-to-end, you may want to build out global stylesheets that allow other application builders across the workflow chain to apply them, saving time and effort. For a stable production app that should match enterprise level branding, create a global stylesheet to ensure that everyone — from builders to end-users — can benefit from consistent styling that strengthens familiarity and accessibility in your Slate application.

If you have a custom style only available in one local application which must remain unique in appearance, build a local stylesheet instead. We'd also recommend sticking with local stylesheets if you're continuously iterating on the look and feel of an application. You can also make that into a global stylesheet when finalized and approved for use across different applications.

For any shape or size of application, style sheets are key to delivering a quality user experience. Get started by reading the Slate stylesheets documentation.

Additional highlights

Analytics: Contour

Allow setting default analysis view | Introduced a new setting to change the default view when opening an analysis. Users can now configure an analysis to open in the dashboard by default or open in the dashboard only if the viewer lacks edit permissions. To change this setting, navigate to the settings sidebar and select an option from the Default open location dropdown menu.

Copy and paste boards in a Contour analysis | Boards can now be copied, cut, and pasted within a single Contour analysis. To copy a board use the ... dropdown menu on any board and select either Copy or Cut. Boards can then be pasted anywhere in the same analysis by using the new Paste board option located in between boards in a path.

---

中文翻译

---

公告

Pipeline Builder：下一代数据集成

发布日期：2022-08-18

旨在实现数据管道的快速、灵活且可扩展交付

大规模构建管道通常是一项成本高昂且受限的工作，需要大量的技术知识、预算和时间投入，才能确保管道健康且全面。由于存在这些基本要求，数据集成对许多用户来说可能难以企及、成本高昂且极难维护。

我们创建了 Pipeline Builder，以便用户可以专注于描述他们想要构建的管道，而无需担心细节。

快速构建管道

Pipeline Builder 为全范围目标输出的批处理、微批处理和流式工作流提供托管式、无服务器支持，同时构建、刷新和其他编排工作可一并大规模处理。

• 无需启动环境或编写样板代码，即可立即对数据（结构化、非结构化、物联网）应用转换。
• 无需对整个数据集进行耗时耗资源的检查，即可在构建失败时立即获得反馈。
• 通过整个管道中细粒度的列溯源，简化调试过程。
• 受益于 Foundry 世界级的安全原语，包括基于目的访问、Palantir 的 Cipher 选择性加密等。
• Pipeline Builder 仅物化满足管道指定目标所需的数据，从而大规模节省计算和存储资源。

精简的数据集成，不牺牲多功能性

Builder 旨在满足任何数据集成需求，内置数百个常用函数，并具备轻松添加新函数的能力。Pipeline Builder 也是后端可扩展的，能够使用不同的执行引擎（例如，用于小规模数据的内存引擎；用于经典批处理管道的 Spark；用于低延迟流式管道的 Flink）。无论底层计算模式如何，Builder 的后端都能确保一致的体验。

此外，现在多个用户可以参与管道构建过程，无需考虑其技术能力。协作和导出为代码是内置的基本功能，有助于构思、创建和审查。如果需要，您可以使用"导出为代码"功能获取管道的 Java Spark 版本。

何时应使用 Pipeline Builder？

Pipeline Builder 使用 Palantir 全面的内置数据转换套件，创建从数据源到最终输出的端到端管道。在集成数据以进行分析和应用构建时使用它。

如果高度定制化的集成需要编写代码，则可以使用 Authoring 通过 SQL、Python 和 Java 对数据应用转换。对于大多数工作流，Pipeline Builder 和 Authoring 相辅相成，Authoring 仓库的输出可作为 Builder 管道的输入，反之亦然。这两个工具都通过版本控制提供协作。集成后，对于分析，我们推荐使用以下工具：

• Code Workbooks： 使用交互式控制台快速迭代转换逻辑、进行临时数据探索、使用常用包（Matplotlib、Plotly、Seaborn）创建详细的交互式图像。
• Quiver： 用于搜索和可视化原生时间序列分析和对象数据。通过可视化点击界面和强大的图表库，在 Ontology 层利用高级分析工作流。Quiver 分析可以模板化为只读仪表板，供更广泛的用户使用。

Pipeline Builder 是一个强大的数据集成工具，可以涵盖管道的端到端创建和管理。要了解有关 Pipeline Builder 的更多信息，请观看我们的发布视频 ↗，或查看我们关于 Pipeline Builder 的文档。

---

Notepad：Foundry 中的现代报告

发布日期：2022-08-15

快速分享洞察并自动化报告生成

Notepad 是一个富文本编辑器，使用户能够分享在 Foundry 平台中生成的洞察，并自动化持续的报告生成。由 Notepad 生成的报告以 Foundry 的 ontology 为后盾，使用户能够快速了解文档中使用的任何数据源的历史和上下文。借助历史沿袭，您的团队可以通过追溯原始输入，轻松解释过去的报告版本或决策。此外，Notepad 文档通过显示使用了相同输入数据的相关或关联内容，进一步帮助连接各个点。

凭借随时可用的上下文和知识，用户能够基于之前的发现构建工作流，而不是完全从头开始。

使用 Notepad 进行下一代报告

Notepad 是一个富文本编辑器，能够嵌入来自整个 Foundry 平台（包括 Contour、Quiver、Foundry Code Workbook 和 Object Explorer）的越来越多的源。Notepad 支持文档模板化，允许您只需单击一下即可创建定期报告的最新版本。

• 嵌入来自其他 Foundry 应用程序的图表、表格和图形。请阅读我们公共文档中关于支持的集成的内容。
• 维护指向嵌入对象和资源的结构化链接，这些链接会自动成为底层 ontology 的一部分。围绕任何给定文档输入的清晰沿袭支持可重现性和可理解性。
• 冻结内容以捕获特定时间点的上下文。
• 模板化文档，以便基于新输入轻松创建报告版本。
• 导出为 PDF 并打印所有 Notepad 文档，并可对导出呈现（嵌入外观、分页符等）进行精细控制。

充分发挥 Notepad 的潜力

Notepad 旨在涵盖从临时报告到模板化定期报告的各种用例：

• 临时笔记： 集成来自其他 Foundry 分析应用程序（如 Quiver）的流程工件和结果。复制并粘贴图表和表格，以实现持久且精确的文档记录。
• 平台内文档： 与平台资源的深度集成支持直接创建文档。
• 定期报告： 使用模板创建蓝图，只需单击一下即可生成最新报告。
• 模板化导出： 文档和嵌入工件针对导出进行了优化。使用模板生成并下载适合导出的 PDF 以进行分享。

Notepad 与 Foundry 应用程序的强大集成以及灵活的编辑选项，帮助其满足您在多种用例中的需求。

何时应使用 Notepad？

使用 Notepad 整合洞察并创建易于构建和消费的文档。考虑您的最终输出，以确定最适合您的 Foundry 生态系统工具：对于基于实时更新和静态文档的大多数报告用例，推荐使用 Notepad。要执行基于用户输入或图表选择的动态过滤（例如，用于展示分析结果的仪表板），请考虑：

• Quiver 仪表板 用于对象或时间序列数据
• Contour 仪表板模式 用于表格数据
• Workshop 用于低代码或无代码以及更高级的操作型应用仪表板

如果您想了解如何无缝地报告和构建集成 Foundry 生态系统中各产品的文档，请参阅 Notepad 文档。

---

Slate：全局样式表

发布日期：2022-08-11

Slate 全局样式表为应用程序构建者提供了一个一站式服务，用于设置多个 Slate 应用程序的样式。

Slate 使用户能够为任何应用程序创建和应用自定义样式，但在您的组织中跨各种应用程序维护一致的样式可能既具挑战性又耗费资源。借助全局样式表，应用程序构建者只需单击一下即可将自定义品牌应用于所有 Slate 应用程序。

一种样式，处处适用

现在，即使有多人构建应用程序，您也可以创建一个紧凑、一致且可识别的视觉品牌。

由于利益相关者通常在工作流中使用多个 Slate 应用程序来深入分析数据，构建者现在可以使用全局样式表在所有应用程序中打造一致的样式，以确保统一的外观和感觉。每当您需要在用户群中引入应用程序一致性和可访问性时，都可以构建全局样式表。

• 多个构建者可以访问为同一组织定制的设计库。
• 选择一个全局样式表，并切换"在应用程序中使用"选项，使同一命名空间中的任何应用程序都采用相同的底层设计指南。
• 以产品支持的方式，将应用程序设计作为一站式流程的一部分进行构建。
• 只需一次操作，即可轻松嵌入可访问的 UI 元素。

为每种所需的视图类型构建全局样式表，以实现混合搭配的适用性

例如，这可能意味着一个用于地图的全局样式表，另一个用于收件箱视图，再一个用于图表或图形。当使用两个或更多相同组件时，可以同时启用多个样式表以组合使用。例如，如果您有一个 Slate 仪表板，您可以应用"仪表板"样式。如果您有一个在暗色模式下查看效果更好的应用程序，并且应用程序构建者已完成该样式，您可以选择"暗色模式"样式。类似地，一个组织的应用程序可以具有相同的品牌和适应性感觉。

要选择设计本地样式表还是全局样式表，请考虑您的样式需求

如果您正在为端到端使用多个应用程序的工作流进行创建，您可能希望构建全局样式表，允许工作流链中的其他应用程序构建者应用它们，从而节省时间和精力。对于应与企业级品牌匹配的稳定生产应用程序，请创建全局样式表，以确保从构建者到最终用户的每个人都能受益于一致的样式，从而增强 Slate 应用程序的熟悉度和可访问性。

如果您有一个仅在一个本地应用程序中可用且必须保持外观独特的自定义样式，请改为构建本地样式表。如果您正在持续迭代应用程序的外观和感觉，我们也建议坚持使用本地样式表。当最终确定并批准跨不同应用程序使用时，您也可以将其制作成全局样式表。

对于任何形状或大小的应用程序，样式表都是提供优质用户体验的关键。请阅读 Slate 样式表文档 开始使用。

其他亮点

分析：Contour

允许设置默认分析视图 | 引入了一项新设置，用于更改打开分析时的默认视图。用户现在可以将分析配置为默认在仪表板中打开，或者仅在查看者缺乏编辑权限时在仪表板中打开。要更改此设置，请导航到设置侧边栏，然后从默认打开位置下拉菜单中选择一个选项。

在 Contour 分析中复制和粘贴面板 | 现在可以在单个 Contour 分析中复制、剪切和粘贴面板。要复制面板，请使用任何面板上的 ... 下拉菜单，然后选择复制或剪切。然后，通过使用路径中面板之间的新粘贴面板选项，可以将面板粘贴到同一分析中的任何位置。