- Dev - mailweb.openeuler.org

【Meeting Notice】sig-ai-bigdata regular meeting Time: 2020-05-21 18:30-20:00
by 正心 17 May '20

17 May '20

1 0

【Meeting Notice】sig-ai-bigdata regular meeting Time: 2020-05-21 18:30-20:00
by Meeting Book 17 May '20

17 May '20

1 0

Upgrade GCC to 9.3.0
by guoge (A) 15 May '20

15 May '20

Hi, dev-team GCC 9.3 has been released on 2020-03-12, and we are ready to upgrade the master branch now. The kernel repo can be compiled through gcc 9.3, no changes needed. Below are the repos that may need to change a little bit, since gcc 9.3 provides libgfortran.so.5, no libgfortran.so.4 provides anymore. atlas future gegl04 gutenprint liberation-fonts liberation-sans-fonts nototools Numpy pcp pygtk2 python-fonttools python-pandas python-pillow python-scikit-learn scipy sil-padauk-fonts suitesparse xsane gimp libvma perl-Archive-Zip perl-MIME-Types the upgrade would be done on May 14th [1], the maintainer of the repos above should look at the daily build report. [1] https://gitee.com/src-openeuler/gcc/pulls/5

3 2

openEuler 能否默认配置 repos 官方源
by mfl 15 May '20

15 May '20

测试在华为云 ECS 通过自定义镜像的方式安装了 openEuler(4.19.90-2003.4.0.0036.oe1.aarch64) ,安装成功后发现 yum ,dnf 默认都没有源配置，建议 openEuler 能在ISO 安装配置默认源。

2 1

回复：答复: sig 组中 Base-service 的 nodejs rmp 包维护者咨询
by xiaoshen＠isrc.iscas.ac.cn 15 May '20

15 May '20

1 0

答复: sig 组中 Base-service 的 nodejs rmp 包维护者咨询
by Hexiaowen (Hex, EulerOS) 14 May '20

14 May '20

胡峰，nodejs找个人接口下xiaoshen 何晓文（Hex EulerOS）庞加莱实验室 [2012实验室] Tel : +86 18042018786 Email : hexiaowen(a)huawei.com [cid:image002.png@01D62A23.A1175090] This e-mail and its attachments contain confidential information from HUAWEI, which is intended only for the person or entity whose address is listed above. Any use of the information contained herein in any way (including, but not limited to, total or partial disclosure, reproduction, or dissemination) by persons other than the intended recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender by phone or email immediately and delete it! 发件人: Zhuyanpeng 发送时间: 2020年5月13日 19:30 收件人: xiaoshen(a)isrc.iscas.ac.cn 抄送: Xiongwei (William, Euler) <xiongwei888(a)huawei.com>; Hexiaowen (Hex, EulerOS) <hexiaowen(a)huawei.com> 主题: 答复: [Dev] sig 组中 Base-service 的 nodejs rmp 包维护者咨询 https://gitee.com/openeuler/community/blob/master/sig/Base-service/OWNERS 发件人: xiaoshen--- via Dev [mailto:dev@openeuler.org] 发送时间: 2020年5月12日 12:23 收件人: dev(a)openeuler.org<mailto:dev@openeuler.org> 主题: [Dev] sig 组中 Base-service 的 nodejs rmp 包维护者咨询您好，我是 sig-cms 组的维护者，我维护的开源项目涉及到 Base-service 的 nodejs，有些问题想咨询一下，您这边能提供一个对接人吗？谢谢！

1 0

Re: sig-ai-bigdata
by 正心 13 May '20

13 May '20

既然这样，那太好了，可以从你最熟悉的地方入手，比如你使用的big data是什么软件栈，可以尝试把你现在用的软件栈(可以具体看)迁移到openeuler上(x86_64和aarch64)，看openeuler上能否满足功能、性能要求，然后我们去弥补差距，这就是很大的贡献。如果你对其他的有兴趣，也可以提出来。 1. 我们的软件库是基于centos还是需要下载源码自己编译? 你说的这些软件库是什么软件库。 2. 我在一个issue里看到您已经跑过spark和TensorFlow。那么最终提供给用户是类似于cloudera这样的集成环境还是原生包让用户选择? 咱们是操作系统，我们要从操作系统的角度思考问题：操作系统不提供这些成熟的商业软件，而是提供基础能力，即当cloudera这样的厂商或使用cloudera的人要在openeuler上使用时，能够很方面的构建、运行起来，功能、性能无损失，甚至有提高。 3. 我是否可以参加下次周会? 是周一还是周五呢？欢迎参加，后面的例会会议通知会提前发到@dev @tc 和sig-ai-bigdata@openeuler.org(筹建中，最近会OK)，所有订阅这三个邮件列表的人都可以收到，并可以接入。祝好。在 2020-05-13 20:48:56，"hao zhang" <unioah(a)gmail.com> 写道：谢谢很详细的讲解。我之前做过一些软件打包，主要是对ubuntu和 archlinux的。我本人一直从事big data方面的开发应用，也算对这一块有些了解。我想我可以先从搭建环境开始， 1. 我们的软件库是基于centos还是需要下载源码自己编译? 2. 我在一个issue里看到您已经跑过spark和TensorFlow。那么最终提供给用户是类似于cloudera这样的集成环境还是原生包让用户选择? 3. 我是否可以参加下次周会? 是周一还是周五呢？谢谢，浩 On Wed, May 13, 2020, 3:44 AM 正心 <sinever(a)126.com> wrote: 浩，你好，非常感谢你的关注，也非常欢迎你的加入。 sig-ai-bigdata的使命如同你在sig的介绍中看到的： 1. 提供openeuler上ai和bigdata的基础运行能力； 2. 提供openeuler上ai和bigdata的竞争力构建； 3. 提供openeuler上ai和bigdata的创新性预研； 4. 提供openeuler上ai和bigdata的新硬件的支持、加速； Openeuler是华为基于上游社区完全从零构建的操作系统，很多软件栈都在构建中，在因此在ai和bigdata上的支持也方兴未艾，当前的首要任务是提供openeuler上ai和bigdata的基础运行能力，包括对各种硬件、各种主流软件包、软件框架的支持，可以做的工作还是很多，如验证典型ai和bigdata软件在openeuler的支持情况，分析他们的依赖，架构梳理，并提供在openeuler的支持(基于openeuler进行构建，提供安装包)，你个人有相关经验吗，如OS打包、ai框架开发使用、bigdata框架的开发使用等或者你个人有什么想法，我们可以讨论或者根据你现有的能力和爱好安排。另外，也欢迎你介绍更多的小伙伴加入进来，一起构建ai和bigdata在openeuler上的未来，期待你的精彩表现。祝 0 error 0 warning sinever | | 正心 | | 邮箱：sinever(a)126.com | 签名由网易邮箱大师定制 On 05/13/2020 00:36, hao zhang wrote: Hi Sinever, 您好，我是新加入到openeuler sig-ai-bigdata 方向的兼职开发人员。根据这次会议的记录，我想在bigdata上面提供帮助。请问具体我可以做些什么？谢谢，浩

1 0

openeuler sig-ai-bigdata第三次例会会议纪要
by 正心 13 May '20

13 May '20

会议时间：2020.05.12 17:30 - 19:00 会议人员：Fred_Li 、cynthia_xh、chxssg、sinever、hubble_zhu、njlzk、zhangrui、xiaoweiwei 会议结论：按照议题：议题一：sig-ai-bigdata前期工作审视与总结 1.完成ai应用视角和系统视角软件栈整理初稿； 2.完成ai领域典型框架tensorflow/caffe在openeuler依赖梳理； 3.完成cml领域典型库libsvm依赖梳理； 4.完成bigdata领域典型应用jupyter依赖梳理； 5.完成引入openeuler支持ai、bigdata技术栈依赖包；议题二：sig-ai-bigdata后期工作内容和方法讨论 1.后续需要继续整理bigdata、ai领域系统视角软件栈，形成全景图； 2.后续从使用角度提供常用软件的常规基本运行能力，如tensorflow/caffe/jupyter/pytorch/hadoop/spark等，提供通常安装方法，如pip、jvm等； 3.对于高性能模式从底层开始逐渐向上支持，如tensorflow支持onednn/cuda/opencl/opencv等； 4.后面引入软件包时先整理使用方法、编译方法、软件架构、依赖关系形成readme，再引入到openeuler；议题三：sig-ai-bigdata后期工作任务分配 1.ai领域软件栈整理，责任人sinever 2.bigdata领域软件栈整理，责任人hubble_zhu、sinever 3.caffe使用方法、软件架构、编译方法、依赖关系整理，责任人njlzk 4.jupyter使用方法、软件架构、编译方法、依赖关系整理，责任人zhangrui 5.hadoop/spark/libsvm使用方法、软件架构、编译方法、依赖关系整理，hubble_zhu 6.其他人任务后续线下视情况再安排；其他： 1.openeuler后续会支持图形界面，因此一些有图形界面的软件也是很需要的； 2.后续例会尽量公开，方面感兴趣的人自行加入，sig-ai-bigdata邮件组建立前可以先使用dev(a)openeuler.com; 3.确认nvidia的相关库是否开源，责任人sinever； 4.后面从x86_64的应用迁移到aarch64可以考虑使用迁移工具； 5.需要整理出sig的roadmap，责任人sinever；

1 0

Roadmap of GCC & OpenJDK in openEuler
by guoge (A) 12 May '20

12 May '20

Hi, tc-team Please review this roadmap of GCC & OpenJDK, two of most important compilers in OS, any questions are welcomed. GCC: The currently supported gcc version is 7.3, it will be upgraded soon, and there will only be ONE existing repo of GCC, the Compiler SIG has prepared well for the next coming gcc 9.3, and it will the candidate(depends on the compatibility & performance between gcc 9.3 and 10.x) default C/C++ compiler in 20.09 LTS. The Compiler SIG has also started the work of gcc 10.1, GNU has released gcc 10.1 on May 7th. OpenJDK: We will support 3 versions of JDK in openEuler，8/LTS/latest，they are all essential components of a commercial OS. 8: Still the most popular adoption of JDK, about over 50% of applications still run on JDK 8[1]. LTS(11): newest version of LTS but incompatible with JDK 8, supporting modularity/AOT/ZGC/stream enhancement/more efficient G1GC/…, we hope developers migrate their code to Java 11 ASAP. Latest(14): OpenJDK is bootstrap but only supports version n-1, which means you can only compile JDK 14 with JDK 13, JDK 15 with JDK 14, and so on, so the latest release of JDK is needed if we want to support the next LTS(17). We have supported OpenJDK 8 & 11 now, they will be updated quarterly to get critical patch update. The next step is to support the latest version of OpenJDK(14), which is under development. [1] https://www.infoq.cn/article/2HFbAm4fDoK6meJfbKW8

1 0

[RM] Release Manager Team Meeting Minutes 20200512
by Maliang (Marlon, IT) 12 May '20

12 May '20

Release Manager Team Meeting Minutes 20200512 Time: 9:15�C10:30 on Tuesday, May 12, 2020 Attendee: Release Manager Team Summary: Marlon Ma Conclusion:�� һ��Sig��&Maintainner��ؼ��ơ��Ͳ��Խ��ȡ��ؼ��Խ��£� Related SIG team and maintainer aligns the progress of key feature Design, Coding, and Testing. The progress of key features is as follows�� n OS��ȷ��1069��һ��601��50%�� OS Dependency Package: 1069 software packages to be supplemented have been sorted out, and the development of the first batch of 601 software packages has been started. The overall progress is 50%. n ��ƽ̨��裺��Ͳ��ƺ�ԭ��֤��25%�� Package Management: Completed software package dependency management and patch tracing, completed design and prototype verification, and started development. The overall progress is 25%. n ��ƽ̨��𡢵��ּ��ɣ��뿪��׶Σ��ݴ��Դ�빹��20%��lkp-test/��Դ��30%�� Community Testing Platform: The deployment and scheduling tasks are completed by level, and the development phase starts. Data processing and source code building: 20%; klp-test/open-source use case design: 30%. n OS��---��ȿ�ܼ��2P MySQL��ݿ�mix��4P MariaDB��ݿ��Ŀ��Ѵ�ɡ��û�̬aware��֪��Ⱥ��ں˹ؼ��ݽṹ�Ż��У��70%�� Domain-based scheduling framework technology: The 2P MySQL database mix scenario and Kunpeng 4P MariaDB database optimization objectives have been achieved. The user-mode aware awareness scheduling and kernel key data structure optimization are in progress. The overall progress is 70%. n OS��---A-Ops��(�ں��滻)��չ40%��Ӧ�ó��ڴ�PIN/�ָ��Լ��ҳ��Ϣ��ɴ��뿪��֤��ɿ��Լӹ̽׶Ρ� A-Ops technology (hot kernel replacement): The overall requirement progress is 40%. The code development for application memory PIN/recovery and restart page information storage has been completed, and the test verification and reliability hardening phases have started. �� ⡣ 1��openEuler��Դ��SPEC�淶�� Review the SPEC specifications of the open-source software package in the openEuler community. Wu Fengguang 2��ˮƽ��Ӧ��Maintainner��Ҫ��ȷ�� The application software maintainer needs to be specified. Ma Liang 3��5.30ǰ��Maintainner��QA Team��ɹؼ��Դ�� ALL Before May 30, the related maintainers and QA team complete the lecture on key features. ALL 4��Ԥ��5�µ�Ready��ҪЭ��һ��Ϲ��Ա�� The Lab required for community development is expected to be ready by the end of May. A material administrator needs to be coordinated. Ma Liang 5��ƽ̨��Ŷ�Crystal��Ի��ϱ��з��ա� �ſ�� The testing platform development team has a weak foundation in the Crystal language, which may cause risks in the short term. Du Kaitian B/R Thanks �� MarlonMa

1 0