I’m often asked, how can I be an IBM Champion for POWER, if I do not own an IBM POWER server or workstation. Yes, life would definitely be easier if I had one. However, I have an over 30 years history with POWER, and there are some fantastic resources available to developers for free. Both help me to stay an active member of the IBM POWER open source community.

Last time I introduced you to the openSUSE Build Service. This time I show you Copr, the Fedora build service.

Copr

Just like OBS, Fedora Copr also started out as a (relatively) simple service to build Fedora and CentOS packages for x86. As Copr is a project by Fedora, the public instance maintained by Fedora at https://copr.fedorainfracloud.org/ only allows you to build open source software. However, you can also install Copr yourself on your own infrastructure. The source code of Copr is available at https://copr.fedorainfracloud.org/, where you can also find links to the documentation.

Today you can use Copr to build packages not just for Fedora x86, but almost all RPM distributions, including openSUSE and OpenMandriva. In addition to x86, you can build packages for 64 bit ARM (aarch64), IBM mainframes (s390x), and IBM POWER 64 bit, little Endian (ppc64le).

You can access Copr using its web interface. There is also a command-line utility, but it was very limited when I last checked. Enabling support for POWER in your project is easy: just select the POWER architecture versions of distributions when you setup the project. You can enable support for POWER also later, but Copr does not automatically build packages for the new architecture. TL;DR: enable support for POWER before building any packages to make your life easier.

How do I use Copr?

Just as with the openSUSE Build Service, my first use of Copr was to make up-to-date syslog-ng packages available to the community. Along the way I used Copr to build some syslog-ng dependencies not yet available in Fedora or RHEL. Some of these are already part of the official distributions.

I did not have a chance yet to benchmark syslog-ng on POWER10, however in the POWER9 era POWER was the best platform to run syslog-ng. I measured syslog-ng collecting over 3 million log messages a second on a POWER9 box when x86 servers could barely go above the 1 million mark.

When I make the latest syslog-ng versions available, I build my EPEL (Extra Packages for Enterprise Linux) packages not just for x86, but also for POWER. I do not know how accurate Copr download statistics are, but for some syslog-ng releases it shows that almost a fourth of all downloads were for POWER syslog-ng packages: https://copr.fedorainfracloud.org/coprs/czanik/syslog-ng44/.

Why Copr?

If your primary focus is to build packages for the Red Hat family of operating systems, Copr provides you with the widest range of possibilities. You can regularly test if your software still compiles on Fedora Rawhide, while providing your users with packages for all the Fedora and RHEL releases. Best of all: even if you do not have a POWER server to work on, you can serve your users with packages built for POWER.

While focused on the openSUSE Innovator initiative as an openSUSE member and Intel Innovator, it was frustrating for me to see that openVINO did not have support on the openSUSE Linux distribution.

In October 2023, I decided to take the personal initiative to start working on compiling and using OpenVINO from the source code for the openSUSE platform. I humbly contributed and published the first adaptations for our distribution on GitHub.

My motivation for this effort stemmed from the potential of OpenVINO to democratize the use of artificial intelligence for those who do not have the resources to invest in expensive GPUs. This library provides multicore programming and the acceleration registers of Intel processors, as well as the resources of ARM processors, allowing the use of AI on processors from the 6th generation onwards.

With the emergence of technologies such as VPU, NPU, and AMX, it is now possible to run LLMs and generative AI without the need for a dedicated GPU. Therefore, I started working on the RPM packaging for openSUSE. This work would not have been successful without the support and assistance of Ilya Lavrenov from Intel and Atri Bhattacharya on the openSUSE Build Service. They not only shared their knowledge with me but also collaborated to ensure compatibility between Intel and openSUSE’s technical policies.

As a result of all this collaborative effort, openSUSE became the first Linux distribution to offer [OpenVINO in its native repository, compiled from the source code. It is a great source of pride to have contributed to this project, which will undoubtedly make a difference in future endeavors. As members of an open-source community, it is our duty to strive to democratize emerging technologies and reduce digital exclusion in society.

For more information, visit here or get it at software.opensuse.org!

A música “Unidos pelo Sul” já está disponível em todas as principais plataformas de streaming de música. Com apenas um simples clique, você não apenas terá a oportunidade ouvir a canção criada 100% por Inteligência Artificial, mas também contribuirá para uma causa nobre, ajudando as vítimas afetadas pelas recentes enchentes. É uma excelente maneira de fazer a diferença na vida de quem precisa.

Spotify:
https://tinyurl.com/rs-spotify-music

YouTube Music:
https://tinyurl.com/rs-youtube-music

Amazon Music:
https://tinyurl.com/rs-amazon-music

Apple Music:
https://tinyurl.com/rs-apple-music

iTunes:
https://tinyurl.com/rs-apple-music

Deezer:
https://tinyurl.com/rs-deezer

TikTok Music:
https://tinyurl.com/rs-tiktok

Neste post apresento a proposta do paper arXiv:2405.03682 , que propõe um fluxo de trabalho que utiliza a Stable Diffusion para aprimorar os resultados de inpainting no contexto de “defurnishing” — a remoção de móveis em imagens panorâmicas internas.

Especificamente, mostra como o aumento do contexto, o ajuste fino do modelo específico para o domínio e a melhoria na mesclagem de imagens podem produzir inpaints de alta fidelidade que são geometricamente plausíveis sem a necessidade de estimar o layout do ambiente. É demonstrado melhorias qualitativas e quantitativas em comparação com outras técnicas de remoção de móveis.

O fluxo de trabalho consiste nos seguintes componentes:

• Pré-processamento: Estimativa de máscaras de móveis por meio de segmentação semântica, rolagem e preenchimento da imagem para garantir um contexto ótimo e redução de resolução para adequação ao pipeline da Stable Diffusion.
• Inpainting: Nosso inpainting personalizado, ajustado para panoramas equiretangulares e resistente a máscaras inexatas e sombras remanescentes, reduzindo assim a tendência da Stable Diffusion em inpainting para criar objetos ilusórios.
• Pós-processamento: Super-resolução e mesclagem das imagens original e inpainted, de modo que os detalhes de alta frequência sejam preservados.

Resultado:

Informações detalhadas do paper: https://arxiv.org/abs/2405.03682

Vuelven las charlas y talleres a Linux Center València, el espacio creado por Slimbook para la divulgación del Software Libre . De esta forma me complace compartir una nueva charla-taller que se celebrará en sus instalaciones que lleva por título «Iniciación a Blender» donde, a cargo de Miguel Menéndez, nos adentraremos en este software de animación en 3D que tan buenos resultados da en este campo.

Iniciación a Blender en Linux Center València

Ya es un clásico del blog hablar de la compañía valenciana de ensamblaje de dispositivos compatibles 100% con el Software Libre conocida como Slimbook. Tras su primera charla en octubre que inició esta nueva temporada de eventos vuelven a promocionar el Software Libre mediante actividades comunitarias que realizan en su Linux Center.

De esta forma el próximo sábado 18 de mayo vamos a poder disfrutar de una nueva charla-taller titulada «Iniciación a Blender» a cargo de Miguel Menéndez , desarrollador que trabaja en el mundo real con Java y Netbeans Platform que tiene varios proyectos relacionados con la programación con Qt tanto en ordenador como en Ubuntu Touch. Además, es autor del Curso de programación de Ubuntu Touch, ha participado en eventos como la Ubuntu Conference 2018 en Gijón, lleva a sus espaldas varios años de contribuciones en foros y comunidades y es el creador y mantenedor de Innerzaurus.

La explicación de la charla/ponencia/taller es el siguiente:

Blender es una aplicación de código abierto y multiplataforma. Permite crear imágenes en 3D de forma sencilla y amigable. Junto al motor 3D, tiene también un motor que permite crear juegos. De esta forma podemos diseñar los modelos del juego en la aplicación y después integrarlos en el juego sin necesidad de programarlo todo desde cero. Este taller es de nivel inicial y en él veremos los pasos para crear nuestro personaje y después añadirlo a un juego.

Para asistir es necesario traer un portátil y tener Blender instalado.

El curso es presencial, por tanto no se va a retransmitir por streaming pero se colgará un resumen del curso en nuestro canal de Youtube, varias semanas después del evento.

La información adicional que necesitas es esta:

Plazas Total: 20 (corre que suelen agotarse)

¿Curso Gratuito o de Pago?: GRATUITO gracias a SLIMBOOK y a Miguel.

Duración del evento: 2h + Preguntas y respuestas

¿Podemos retransmitirlo via streaming?: No

Localización: Linux Center (Grupo Odín). Ronda de la Química s/n Edificio ABM L’Andana, 7ª planta Frente a Parque Técnológico 46980 Paterna, Valencia

Más información: Linux Center

La entrada Iniciación a Blender en Linux Center València se publicó primero en KDE Blog.

Me congratula presentaros el episodio 27 de KDE Express, titulado «Programa #akademyes València 2024» donde de David Marzal y Baltasar Ortega (un servidor), con el apoyo de Jorge Lama en la post-producción hacemos un repaso de lo que nos espera en el mayor evento kdeero de este año en España

Episodio 27 de KDE Express: Programa #akademyes València 2024 esLibre Edition

Comenté hace ya bastante tiempo que había nacido KDE Express, un audio con noticias y la actualidad de la Comunidad KDE y del Software Libre con un formato breve (menos de 30 minutos) que complementan los que ya generaba la Comunidad de KDE España, aunque ahora estamos tomándonos un tiempo de respiro por diversos motivos, con sus ya veteranos Vídeo-Podcast que todavía podéis encontrar en Archive.org, Youtube, Ivoox, Spotify y Apple Podcast.

De esta forma, a lo largo ya casi los 30 episodios, promovidos principalmente por David Marzal, nos han contado un poco de todo: noticias, proyectos, eventos, etc., convirtiéndose (al menos para mi) uno de los podcast favoritos que me suelo encontrar en mi reproductor audio.

En palabras de David el nuevo episodio de KDE Express número 27 los siguientes temas:

Episodio monografico sobre el programa de charlas y actividades de Akademy-es 2024 Valencia.

Os detallamos lo que podeís encontrar en https://www.kde-espana.org/programa-akademy-es-2024 y recordamos que teneis varios métodos de contacto para estar informados en tiempo real los días del evento en https://kdeexpress.gitlab.io/contacto/

¡Nos vemos en la Akademy-es Valencia el 24 y 25 de mayo!

Gracias por el apoyo a openSUSE, La Universidad de La Laguna y a todos los que vengáis o compartáis.

Agradecimientos: Jorge Lama por su asistencia, consejo, apoyo y edición de audio en este episodio.

Y, como siempre, os dejo aquí el listado de los episodios. ¡Disfrutad!

Por cierto, también podéis encontrarlos en Telegram: https://t.me/KDEexpress

La entrada Episodio 27 de KDE Express: Programa #akademyes València 2024 esLibre Edition se publicó primero en KDE Blog.

Me complace promocionar un nuevo audio de Mancomún Podcast que trata de dar visibilidad a uno de las reuniones más importantes de las Comunidades del Software Libre de España. De esta forma os presento la participación de Akademy-es 2024 de València #esLibre edition en Mancomún Podcast otro episodio presentado por Jorge Lama y con la presencia de David Marzal y un servidor, Baltasar Ortega.

Akademy-es 2024 de València #esLibre editionen Mancomún Podcast #akademyes

Uno de los podcast que más me gustan y que suelen durar poco tiempo en mi lista de reproducción es Mancomún Podcast. Sus audios presentados habitualmente por Jorge Lama y Brais Arias (miembro de KDE España) suelen ir directos y al grano, con lo que casi nunca superan la hora de duración: ideal para conocer nuevos proyectos, iniciativa y personas.

El episodio publicado el pasado 9 de mayo estubo dedicado a Akademy-es 2024 esLibre Edition. En palabras de Jorge:

Enlaces mencionados no programa Akademy-es 2024:
https://www.kdeblog.com/programa-de-charlas-de-akademy-es-2024-de-valencia-eslibre-edition-akademyes.html
eslibre 2024:
https://eslib.re/2024/

Aprovecho para poner enlaces generales del proyecto de Mancomún.

Repositorio de la nueva temporada de Mancomún: https://archive.org/details/mancomun-podcast-101

Si os queréis subscribir este es el feed: http://feeds.feedburner.com/MancomunPodcast

Más información: Mancomún Podcast

La entrada Akademy-es 2024 de València #esLibre edition en Mancomún Podcast #akademyes se publicó primero en KDE Blog.

Como los lectores habituales del blog sabrán, el pasado 28 de febrero la Comunidad KDE realizó un importante salto tecnológico, uno que va a marcar su evolución para los próximos años. Este gran cambio a las librerías Qt 6 nos proporcionó el nuevo escritorio Plasma 6, del que ya he hablado a lo largo de muchas entradas de marzo, y el ecosistema de aplicaciones KDE Gear 24.02, del cual desgrané a lo largo del pasado mes de abril. Pero no solo fue eso, sino que además nos trajo el salto también a KDE Framworks 6, las librerías propias del proyecto KDE. Tantas noticias han aparecido de los anteriores proyectos que no he tenido tiempo todavía de hablar de esto, y lo hago hoy dado que ayer se anunció la segunda actualización de KDE Frameworks 6. Es el momento de dar su parte de protagonismo.

Segunda actualización de KDE Frameworks

A pesar de que para los usuarios corrientes esta noticia sea algo confusa ya que no se trata de realzar una nueva aplicación ni de una nueva gran funcionalidad del escritorio, el desarrollo de KDE Frameworks tiene repercusiones directas en él a medio y largo plazo.

Para los que no lo sepan, KDE Frameworks añade más de 83 librerías a la propias de Qt que proporcionan una gran variedad de funcionalidades necesarias y comunes, precisadas por los desarrolladores, testeadas por aplicaciones especí­ficas y publicadas bajo licencias flexibles.

De esta forma, KDE Frameworks se convierte en la base de trabajo de los desarrolladores para realizar sus aplicaciones o sus desarrollos para los entornos de trabajo (escritorio para ordenadores, plasma mobile, etc).

Un buen símil es que KDE Framworks es como el papel y las herramientas de dibujo para un artista: cuanto mejor sea el papel y mejores pinceles tenga, la creación de una artista será mejor.

Como he dicho, el pasado 28 de febrero KDE Frameworks saltó de la versión 5 a la 6, y ha sido hoy viernes 10 de mayo cuando se ha anunciado que ya tenemos la segunda actualización de la rama, es decir, que ha sido lanzado KDE Frameworks 6.2.

Hay que destacar que esta versión forma parte de una serie de versiones mensuales planificadas para poner las mejoras a disposición de los desarrolladores de forma rápida y previsible y que es absolutamente recomendable su actualización.

Aquí podéis encontrar un listado con todos estos frameworks y la serie de artículos que dedico a KDE Frameworks en el blog,

Más información: KDE

La entrada Segunda actualización de KDE Frameworks se publicó primero en KDE Blog.

A IBM introduziu ao mercado a Granite, uma nova série de modelos de inteligência artificial (IA) de código aberto, destinados a facilitar o processo de codificação para desenvolvedores em diversos setores. Os modelos, disponíveis em plataformas como Hugging Face, GitHub, watsonx.ai, e RHEL AI, são liberados sob a licença Apache 2.0. Eles foram projetados para auxiliar desenvolvedores na escrita, teste, depuração e distribuição de software confiável.

Granite apresenta quatro variantes, que variam de acordo com o tamanho do banco de dados, de 3 a 34 bilhões de parâmetros, de acordo com o Analytics India Magazine. Os modelos foram submetidos a testes rigorosos em diversos benchmarks e se mostraram superiores a outros modelos de código aberto semelhantes, como Code Llama e Llama 3, em várias tarefas de programação. Essa superioridade é atribuída ao seu treinamento em um vasto conjunto de dados composto por 500 milhões de linhas de código abrangendo mais de 50 linguagens de programação, permitindo-lhes identificar padrões e solucionar bugs complexos em linguagens como Python, JavaScript, Java, entre outras.

Além de suas aplicações em geração de código e testes, os modelos Granite são também adequados para automatizar tarefas rotineiras, tais como a criação de testes unitários e a elaboração de documentação técnica. “Acreditamos no poder da inovação aberta e queremos alcançar o maior número possível de desenvolvedores”, disse Ruchir Puri, cientista-chefe da IBM Research, expressando entusiasmo quanto ao potencial de desenvolvimento de novas ferramentas e software com o uso de Granite.

Informações: https://github.com/ibm-granite/granite-code-models

A Shallow Understanding of openSUSE

Introduction

As part of my job at SUSE, I have been using openSUSE Leap and Tumbleweed, in addition to SLE Micro for a while now. After some experimentation last weekend, and asking some friends some questions, I think I finally have a mental model of SUSE's distro options. I thought I would test that understanding by trying to write it out a bit here. I already know that there are a lot of details I glossed over, and a lot of people see things differently.

I think that openSUSE is the most useful and stable distro available, and more people should know about it and use it. Part of what makes SUSE’s distros great is the community and open source ethics that go into the distros, including the Enterprise versions. With any healthy open source community comes innovation, and with innovation comes choice. But with choice comes decision making, it can be hard to know where to start.

In general, when choosing a SUSE distro, I think there are 3 points to consider:

1. What package lifecycle do you want?
2. How do you want to manage the OS?
3. Do you want Enterprise support or not?

Lifecycle

openSUSE has two major flavors of distros: Tumbleweed and Leap.

Tumbleweed, The Reliable Rolling Release

The major difference between them is that Tumbleweed’s package repositories are constantly updated, and as soon as the automated tests all pass, package updates are released to users. Major and minor version updates can be released as soon as they are ready and tests pass, so distribution updates are available as often as daily.

Unlike rolling releases from other distros, Tumbleweed updates are low drama for the user. They almost always work, and the packages are securely built on SUSE’s Open Build Service, so there is no fussing with local compilation and such.

Tumbleweed provides a great option if you like to have the most up to date software, but you don’t want to hassle with keeping the software up to date yourself. I am running it on a laptop that I use for work, and it is working quite well.

Leap, The Stable Release

By contrast, Leap tracks the SUSE Linux Enterprise release cycles. The package versions in the repositories receive security and bug fixes, but new versions of the software are only provided via point releases. For example, the upcoming release of 15.6 will offer some updated package versions.

This means that you can lock in on a working configuration and run it for a long time with minimal maintenance and overhead.

“Stability”

Both Tumbleweed and Leap are “stable” in the sense that they pass quality control checks and aren’t prone to crashes etc… I like to use the word “reliable” for this quality of openSUSE. By “stability” in this context, I really refer to how often the major package versions do or do not change. Management

Management

The second dimension that you can consider is if you want to manage a read only file system, or a traditional file system. This boils down to deciding between traditional openSUSE and Micro.

Both versions support transactional updates, as described below.

Traditional

If you opt for a traditional install (as I do) then you manage the system with zypper (or whatever package management tool you prefer) and use zypper to install packages like applications, and to keep the system up to date. Zypper makes changes directly to the system by doing things like installing dependencies, etc…

If you are using a traditional SUSE flavor, you don’t typically need to reboot the system after running installations and updates (there are exceptions of course). You can also use repositories to install software packaged as RPMs to install directly on the system or use things like Flatpaks and containers, depending on your preference.

Before updating the system or installing software, it’s a good idea to take a snapshot so that you can roll back easily if you need to or want to. This is done with the tool Snapper. One neat thing is that, in the default configuration, when you run the zypper dup command, it takes a snapshot for you automatically.

Micro

“SUSE Linux Micro” is the Enterprise version, whereas “openSUSE Leap Micro” is the community Leap version, and “openSUSE MicroOS” is the community Tumbleweed version. For simplicity I will refer to them collectively as “Micro” here.

Micro flavors come with a read only root file system. This provides a lot of benefits in terms of safety, but it requires a different approach to managing the system.

If you try to use zypper to install, it won’t work because you can’t make changes directly to the filesystem. Rather, if you want to install an RPM, you need to create a transaction wherein Micro will create a new copy of the root filesystem, update that copy, and then reboot into it. If the reboot fails, no problem, it will fall back to the original. As such, you can’t really use zypper in the normal way, but rather apply transactional updates.

Micro is designed with Flatpak, Containers, and VMs in mind for running applications. The idea is that your applications are containerized and can be safely installed and updated without touching the root filesystem.

In this approach, your applications come with their dependencies bundled in, so there are fewer dependencies in the base system in the first place. This makes Micro smaller, but also means that there are fewer dependencies to be updated, so updating the base system is less frequent as well.

Micro also keeps a list of previous states, so you can easily rollback to a previous transaction point for the base OS.

Enterprise Support

SUSE’s Enterprise Support includes critical benefits such as:

Certifications such as CC EAL 4+, FIPS, etc… Dedicated support engineers available around the clock Level 3 Support Dependable, clearly documented, and long life cycles for releases

If you need, or think you might need to opt for Enterprise support but you want to start with openSUSE, then you should stick with Leap (either traditional or Micro). This is because Leap is built out of the same repositories that SUSE Linux Enterprise is built out of, and opting into SUSE Support can, in some cases, be as simple as installing some packages.

Conclusion

Like any healthy distro, the openSUSE community has seen a flowering of innovation as community members put together solutions based on their own interests. For example, the community also maintains a sort of “in between” version called Slowroll. There is also an image based on MicroOS that includes a full desktop experience, called Aeon, and lots of other options. Too much richness to cover here, but I tried to cover the essential decision points.

Want some links? Community supported “choose your version” site

Leap 15.5 (stable traditional version
Tumbleweed (rolling traditional version
openSUSE Leap Micro (stable read only file system version
MicroOS (rolling read only file system version