This week openSUSE Tumbleweed snapshots continue to deliver software updates for users wanting to enhance their systems.

Updates in this week’s snapshots brought numerous changes, bug fixes and improved functionality.

With snapshot 20230926, the 2D graphics library cairo had its first stable release in five years with the 1.18.0 version update. The package includes support for Type 3 color fonts in PDF and has several optimizations including enhancing support for the Quartz surface and the removal of the XML surface. The snapshot also updates image editor ImageMagick; the 7.1.1.17 version modifies patches and introduces alternative configuration packages, eliminating a redundant dependency on p7zip. An update of libreoffice 7.6.1.2 resolves a crash issue when editing a date field, fixes animation effects for rectangles and fixes an accuracy problem with SUM() calculations. The new 4.4.0 version of transactional-update has a new setup-fips command and improved cleanup processes for snapshots created through various methods. Previously, only snapshots created by the transactional-update shell script were cleaned up. The update of yast2-bootloader 5.0.2 fixes issues in the testsuite for architectures other than x86_64 and introduces support for systemd-boot on x86_64, which can be enabled/disabled by the enable_systemd_boot flag in the product description file for each product. The update of yast2-installation 5.0.1 accommodates those changes in yast2-bootloader for systemd-boot experimental support. Several other packages were updated in the snapshot.

An update in snapshot 20230925 includes a package to make computer programs run faster; the highway 1.0.7 package introduces new features like LoadNOr, GatherIndexN, and ScatterIndexN. The new version also had fixes for WASM, GNU Compiler Collection 13, GCC 8.2, QEMU 7.2 and more. The GObject library interfacing with system block devices, udisks2, updates to version 2.10.0 and has improvements with changes to partitioning using libfdisk and has support for native NVMe devices. The package also enhances the handling of filesystems and adds new features like setting filesystem and partition UUIDs. The libHX 4.15 update introduces new functions for computing the Least Positive Residue and enhances the HX_strrtrim function to work with strings longer to improve its string manipulation capabilities. An update of the Library for manipulating block devices libblockdev had a major verison update to 3.0.2. The package improvements include using ntfsinfo for faster operations, brings extensive Application Programming Interface changes, removes certain plugins, and has new features like NVMe support. A few other packages updated in the snapshot.

An update of bind 9.18.19 arrived in snapshot 20230922. The update of this DNS protocol package fixes a few Common Vulnerabilities and Exposures. Both CVE-2023-3341 and CVE-2023-4236 were fixed and the latter could occurs when handling DNS-over-TLS queries are under heavy load. This could have caused internal data structures to be incorrectly reused and potentially has an impact on the stability and reliability of the DNS server. The update of gtk4 4.12.2 prevents tooltips from crossing native boundaries in GtkTooltip and improves gradient matching and rendering in GSK. An update of the HP Linux Imaging and Printing package, hplip moved to version 3.23.5 and now supports several new printers and models like the HP Smart Tank series. An update of xwayland 23.2.1 provides a more stable and functional experience while adding a check to verify the existence of a function call. An update of Xen 4.17.2 implements two security fixes. CVE-2023-34322 addresses a security concern where the top-level shadow reference for 64-bit PV guests was dropped prematurely when dealing with memory shortages and may have lead to privilege escalation, Denial of Service (DoS), and information leaks. CVE-2023-20588 was the other vulnerability. Several other packages updated in the snapshot including ncurses 6.4.20230909, nodejs20 20.7.0, python-Pillow 10.0.1 and more.

With snapshot 20230921, ethtool 6.5 adds a feature to check the register dump for a specific driver and fixes a few issues like some problems with certain settings and an empty slot search issue. The iproute2 6.5 package, when using the rdma command will report the device protocol, which can be handy. When using iplink, it doesn’t use all the options provided and can give an error message. The Linux Kernel updates in the snapshot as kernel-source 6.5.4 improves the handling Server Message Block 3 requests after reconnects, has driver fixes and enhancements related to clocks and drivers for various hardware components. The update of lvm2 2.03.22 fixes report field types for better data accuracy and enhancing Virtual Data Optimizer and ensures more efficient data management in Linux storage systems. The audio package pipewire updates to 0.3.80 and improves better synchronization of mixer I/O areas to prevent potential crashes. The update also has Advanced Linux Sound Architecture making smarter period-size selections for batch devices and provides enhancements in audio adapter tag and latency handling. PDF render poppler 23.09.0 improves digital signatures and now provides the key location for certificates to be signed.

Shortly after releasing last week’s review, snapshot 20230920 was released and it brought GNOME users a gift for the weekend. GNOME 45 brings several changes. A new activities button in the top bar has a dynamic workspace indicator, making it more user-friendly and replacing the old app menu. There is faster search performance across various apps and a new camera indicator in the top bar shows when the camera is in use. GNOME 45 introduces a new default image viewer that’s fast, clean and adaptive and the files app is now faster and more seamless with search. The GNOME Display Manager gdm 45.0.1 fixes a critical crash issue. Has numerous code cleanups and adds support for multi-seat configurations in Wayland. An update of Flatpak 1.15.4 was also in the snapshot. The package took care of two CVEs. CVE-2023-28101 has a fix for special characters that are now escaped when displaying permissions and metadata. This prevents malicious apps from manipulating the appearance of the permissions list using crafted metadata. With CVE-2023-28100, Flatpak app is run on a Linux virtual console and ioctl is no longer allowed. This security enhancement protects against potential vulnerabilities when using virtual consoles. There were several other packages updating in the snapshot, but the major version of postgresql 16 optimizes aggregate functions, and enhances sorting that leads to faster queries. The bulk loading performance saw up to a 300 percent improvement, according to its release announcement. The other major version in the snapshot was gedit 46.1, which resolved a build issue, updated its Flatpak and did some code refactorings to improve organization.

No dia 25 de setembro de 2023, a OpenAI revelou a introdução de dois recursos inovadores que expandem as maneiras com as quais as pessoas podem interagir com seu modelo mais avançado e recente, o GPT-4: a capacidade de questionar sobre imagens e de usar a voz como entrada para uma consulta.

Neste Post, as primeiras impressões do pessoal da Roboflow com o recurso de entrada de imagem do GPT-4V. Foi conduzido uma série de experimentos para averiguar as funcionalidades do GPT-4V, demonstrando os pontos onde o modelo se sai bem e onde enfrenta obstáculos.

O que é GPT-4V?

GPT-4V(ision) (GPT-4V) é um modelo multimodal criado pela OpenAI. Ele permite que os usuários façam o upload de uma imagem e realizem perguntas sobre ela, uma tarefa denominada como resposta a perguntas visuais (VQA).

O lançamento do GPT-4V começou em 24 de setembro e ele estára acessível tanto no aplicativo OpenAI ChatGPT para iOS quanto na plataforma web.

Foram diversos testes por parte da equipe Roboflow, mas o que mais me impressionou foram as habilidades do GPT-4V em responder perguntas ao questionarmos sobre um local específico. Foi Submetido uma foto de São Francisco juntamente com a pergunta “Onde é isso?”.

O GPT-4V reconheceu acertadamente o local como São Francisco e destacou que a Pirâmide Transamerica, presente na foto enviada, é um marco icônico da cidade.

Outro teste impressionante foi a capacidade de OCR do GPT-4V: OCR em uma imagem com texto em um pneu de carro. A intenção era construir um entendimento de como o GPT-4V se comporta em OCR em cenários reais, onde o texto pode ter menos contraste e estar em um ângulo não perpendicular.

O GPT-4V não conseguiu identificar corretamente o número de série em uma imagem de um pneu. Alguns números estavam corretos, mas houve vários erros no resultado fornecido pelo modelo.

OCR Matemático é uma versão especializada de OCR destinada especificamente a equações matemáticas. Ele é muitas vezes considerado uma disciplina própria, dado que a sintaxe que o modelo de OCR necessita reconhecer abrange uma ampla variedade de símbolos.

Submetemos uma questão de matemática ao GPT-4V. Esta pergunta estava contida em uma captura de tela de um documento. A pergunta refere-se ao cálculo do comprimento de um fio de tirolesa, dados dois ângulos. Apresentamos a imagem com a instrução “Resolva isso.”

O modelo reconheceu que o problema poderia ser solucionado através de trigonometria, determinou a função apropriada a ser utilizada e demonstrou, passo a passo, como solucionar o problema. Em seguida, o GPT-4V apresentou a resposta correta para a questão.

GPT-4V em Visão Computacional e Mais

O GPT-4V representa um avanço significativo no campo da aprendizagem de máquinas e processamento de linguagem natural. Com ele, é possível fazer perguntas sobre uma imagem – e suas respectivas perguntas de seguimento – de maneira natural, e o modelo tentará fornecer uma resposta.

O GPT-4V demonstrou bom desempenho em diversas perguntas genéricas sobre imagens e mostrou ter consciência de contexto em alguns dos casos que testamos. Por exemplo, conseguiu responder corretamente a perguntas sobre um filme retratado em uma imagem sem que lhe fosse dito textualmente qual era o filme.

O GPT-4V é promissor para responder perguntas gerais. Modelos anteriores, destinados a este propósito, frequentemente apresentavam respostas menos fluentes. O GPT-4V, por outro lado, é capaz de responder a perguntas e suas respectivas de seguimento sobre uma imagem com profundidade.

Com o GPT-4V, perguntas sobre uma imagem podem ser feitas sem necessidade de um processo em duas etapas (isto é, classificação e, em seguida, usar os resultados para questionar um modelo de linguagem como o GPT). É provável que existam limitações no que o GPT-4V pode compreender, assim, testar um caso de uso para entender o desempenho do modelo é fundamental.

Com isso dito, o GPT-4V possui suas limitações. O modelo “alucinou”, fornecendo informações incorretas. Isso representa um risco ao utilizar modelos de linguagem para responder perguntas. Ademais, o modelo não conseguiu delimitar objetos com precisão para detecção, indicando que, atualmente, não é adequado para este caso de uso.

Também notamos que o GPT-4V não responde perguntas sobre pessoas. Quando apresentado com uma foto de Taylor Swift e questionado sobre quem estava na imagem, o modelo optou por não responder. A OpenAI considera isso um comportamento previsto, conforme descrito no cartão de sistema publicado.

https://openai.com/research/gpt-4v-system-card

You can read about many interesting syslog-ng features in my blogs. However, it can happen that when you want to try them at home, you fail because the feature is missing. How can you solve such problems? In this blog, I discuss some of the possible solutions from installing sub-packages through using unofficial repositories, to upgrading your OS.

This blog focuses on RPM packages for openSUSE / SLES, Fedora / RHEL, and FreeBSD, because these are the packages I know – I am their maintainer. However, these problems and their solutions also apply to Debian / Ubuntu, and other Linux distributions.

https://www.syslog-ng.com/community/b/blog/posts/why-is-a-feature-not-available-in-the-syslog-ng-package

Me complace compartir con vosotros el episodios de septiembre de 2023 de GNU/Linux València en el que hablan un poco de todo como la utilización de un sextante de rayos cósmicos como alternativa al GPS o que Linux pasa a MacOS en Steam:.

Episodio de septiembre de 2023 los podcast de GNU/Linux València

Tras bastante tiempo en silencio, básicamente porque no tenían forma de cuadrar agendas, han vuelto con mucha fuerza los podcast de GNU/Linux València. Unos podcast que suelen ser de tipo tertulia muy amena y destinada a todo el mundo.

Podcast del 27 de septiembre

El primer podcast después de vacaciones ya está aquí. Grabado el 3 de septiembre y lleva por título «Steam, freeBSD y más». Una serie de noticias que destacan porque al final todo acaba en que Linux está en todos los sitios menos en los ordenadores de sobremesa o portátiles de los usuarios. En fin.

Más información: Podcast 27 de septiembnre de GNU/Linux València

¡Únete a GNU/Linux València!

Aprovecho para recordar que desde hace unos meses, los chicos de GNU/Linux Valencia ya tienen su menú propio en el blog, con lo que seguir sus eventos en esta humilde bitácora será más fácil que nunca, y así podréis comprobar su alto nivel de actividades que realizan que destacan por su variedad.

Y que además, GNU/Linux València creció y se ha convertió en asociación. Así que si buscas una forma de colaborar con el Software Libre, esta asociación puede ser tu sitio. ¡Te esperamos!

La entrada Episodio de septiembre de 2023 los podcast de GNU/Linux València se publicó primero en KDE Blog.

En pocos meses vamos a tener un nuevo lanzamiento de un escritorio de la Comunidad KDE. No será un lanzamiento menor, sino todo un lavado de cara del entorno de trabajo que da el salto a la tecnologías que ofrece Qt 6. Es por ello que se ha decidio empezar una campaña de recaudación de fondos para Plasma 6 que dote de músculo económico a la Comunidad para afrontar el gran reto que supone una actualización de este tipo. Vienen meses de trabajo, reuniones, eventos y sprints para dejarlo pulido.

Campaña de recaudación de fondos para Plasma 6

Donar a proyectos abiertos es una de las formas de colaborar más sencillas en el desarrollo de estas iniciativas. Y es que debemos tener claro que gran parte del Software Libre es gratuito pero implica unos gastos básicos, por ejemplo, en cuanto a servidores. Además, si queremos calidad y que se tenga garantías de desarrollo los usuarios, o gran parte de estos, deben colaborar de alguna forma en su desarrollo. Creo que es un buen momento para invitaros a leer este artículo ya algo viejo que publiqué en su día en este humilde blog.

De esta forma se inicia una campaña de recaudación de fondos para KDE para que la Comunidad KDE siga creciendo y tenga los recursos necesarios para afrontar los posibles retos que seguro aparecen. Además, de esta forma se comprometen a serguir ofreciendo más y más software de calidad de forma gratuita, tanto de forma monetaria como en forma de datos personales.

El espiritu con el que se inicia esta campaña es el siguiente:

¡Noticias emocionantes en el horizonte! En febrero de 2024, Plasma 6 hará su gran debut. Pero necesitamos tu ayuda para garantizar el éxito de su lanzamiento.

Y es que la Comunidad KDE necesita dinera para optimizar actividades como:

• Sprints para desarrolladores: Ayudarás a financiar las reuniones presenciales que mantienen a nuestros desarrolladores con energía y centrados en hacer KDE aún mejor.
• Gastos de viaje a eventos: Apoyarás la presencia de nuestro equipo en reuniones y conferencias importantes, como FOSDEM, FOSSAsia y LinuxCons.
• Evento Akademy: Garantizará el éxito del evento anual de la comunidad KDE para todos los miembros, y fomentará la colaboración y el crecimiento.
• Funcionamiento de KDE: Mantendrás las luces encendidas en la sede de KDE y nuestro hogar digital funcionando sin problemas.
• Pagar al personal de apoyo: Te asegurarás de que KDE tenga a mano los expertos que necesitamos para ayudar a nuestros colaboradores y usuarios.

En palabras de los desarrolladores, la razón principal por la que deberías donar es:

Porque la generosidad está en el corazón de KDE y mantiene vivos nuestros proyectos.

Es lo que motiva a nuestros voluntarios a compartir su tiempo y conocimientos para seguir construyendo nuevas características y mantener el software de KDE actualizado y seguro.

La generosidad también impulsa a quienes nos apoyan. Sus contribuciones y patrocinios hacen que los engranajes de KDE sigan girando y garantizan que nuestros desarrolladores puedan continuar con su fantástico trabajo.

Además tu donación te ofrece un reconocimiento ya que, a esperas de alguna que otra sorpresa que se uniará a los siguientes ítems:

• Tu nombre aparecerá en nuestra página de donaciones, reconociendo su contribución.
• Tu nombre aparecerá en el propio Plasma 6.

Si no quieres nada de lo anterior, también está bien, por supuesto. Recuerda marcar la casilla «[✔️] Hacer donación anónima» en el proceso de donación anterior.

Y si eres un desarrollador, puedes beneficiarte de las herramientas y marcos de trabajo de KDE. Las bibliotecas de KDE hacen que sea rápido y fácil construir aplicaciones sofisticadas para todas las plataformas.

Más información: KDE

La entrada Campaña de recaudación de fondos para Plasma 6 se publicó primero en KDE Blog.

A new version of the modern lightweight host operating system Leap Micro 5.5 just entered the Beta of its development.

A quick transition to a Release Candidate (RC) expected and the General Availability (GA) slated for the first half of October.

One of the standout features of Leap Micro 5.5 is its SELinux enhancements. Security-Enhanced Linux (SELinux) has received a significant boost; It brings podman-docker and hyper-v support for AArch64 for a more robust and secure computing experience for users.

In tandem with this exciting update, it’s important to note that Leap Micro 5.3 is nearing its End of Life (EOL). As soon as Leap Micro 5.5 makes its official debut, Leap Micro 5.3 will be retired. Users of Leap Micro 5.3 are strongly advised to consider upgrading to either Leap Micro 5.4 or the upcoming 5.5 release. This ensures access to the latest features, security enhancements, and ongoing support.

Este 27 de septiembre de 2023 se cumplen 40 años el anuncio inicial del proyecto GNU realizado por Richard Stallman

Richard Stallman anunció un 27 de septiembre de 1983 que empezaría a trabajar en un sistema operativo llamado GNU similar a Unix pero con la principal diferencia que sería completamente libre para usarlo, estudiarlo o modificarlo y distribuir esas modificaciones.

Puede determinarse este el inicio del movimiento del software libre, que en 40 años ha permeado en multitud de personas y proyectos.

Cuando se celebraron los 30 años del proyecto GNU escribí un artículo en el blog, que bien podría servir para conmemorar este nuevo aniversario. Muchas cosas han ocurrido en estos 10 años en el mundo del software libre y muchos cambios han ocurrido.

Pero en este caso, no quería hablar solo de fechas, o poner enlaces. Aunque sí enlazaré a la web oficial donde se detalla el evento que tendrá lugar en Suiza con charlas, entre ellas de Richard Stallman o Luis Falcón de GNU Health y otras charlas relacionadas con el software libre.

Pero en esta ocasión quiero hablar específicamente sobre mi visión del software libre.

Lo he dicho en varias ocasiones y sigo creyendo que es cierto. El software libre es algo más que simples bits o licencias que permitan que el software pueda ser usado, estudiado o modificado por otras personas.

Creo en el componente ético que subyace debajo de las funciones y líneas de código de cualquier programa que cumpla las 4 libertades del software libre.

En estos 40 años de existencia el software libre ha hecho posible y ha dado sustento legal y moral a muchos proyectos hoy en día indispensables. El software libre ha dado las herramientas y el pretexto para que multitud de mentes se pudieran unir y pusieran en marcha un proyecto nuevo, continuaran alguno ya acabado o dieran una alternativa a software con una licencia privativa que ata.

Así pues podemos encontrar desde pequeños proyectos específicos hasta grandes proyectos más generales y de un público más amplio que se imbuyen de ese espíritu del software libre que hace que se aglutinen personas de todo el mundo entorno al proyecto, ya sea aportando código, traduciendo, probando en distintas plataformas, etc.

El software privativo planteó su modelo de negocio en el código. Se vendía el código. Después vieron que teniendo atadas a las personas que utilizaban su código era una buena idea explotar los datos y metadatos que generaba el uso de ese código cerrado.

Una de las constantes dentro del software libre es el respeto a la persona que lo utiliza. El tener expuesto el código que se crea, hace imposible crear código malicioso que espíe o sea intrusivo con la privacidad. Siempre habrá alguien que se lea el código fuente y exponga, si lo hubiera, esa funcionalidad no deseada.

En estos 40 años, el software libre ha crecido y se ha expandido en muchos aspectos y proyectos. Las empresas han visto ese potencial y han “desvirtualizado» ese software libre por un término más “suave» que no implique radicalidad y ha aparecido el “open source» o código abierto.

Sin duda que el código sea abierto es una de las premisas del software libre, pero por eso solo no basta. Y además sigo pensando que sigue siendo muy necesario ese componente ético que aporta el software libre.

El tema de las licencias también se ha complicado. Las licencias de software libre implican que quien use ese software para crear algo nuevo, deba publicarlo a su vez bajo una licencia de software libre.

No han sido pocas las ocasiones en las que empresas utilizan licencias permisivas, pero no software libre y después han cerrado esa licencia. Eso con licencias de software libre no pasa.

Seguiré llamándolo como tal software libre, frente a “open source» o similar. Seguiré apoyando las opciones de software libre. No soy perfecto (alguien lo es) e incurriré en incoherencias, pero trataré de que sean las mínimas o de aprender de quien me señale el error.

Felicitar a Stallman como pionero de este proyecto del software libre y hacer extensiva esa felicitación a todas las personas que de una u otra manera crean y fomentan software libre, ya sea de manera profesional en su trabajo diario o de manera altruista contribuyendo con diferentes proyectos.

Empecé en GNU/Linux sin tener ni idea de qué era eso del software libre y poco a poco, leyendo y aprendiendo lo fui conociendo. Hoy reclamo ese derecho a poder utilizar software libre en la medida de lo posible en mis dispositivos. Y eso lo hacen posible un montón de gente en todo el mundo.

Happy hacking!

A recently published openSUSE survey is asking IT professionals and users about their views on open-source technologies and the ever-evolving Linux ecosystem.

The Use Case Survey aims to gather insights on what will shape the future of Linux and its role in various industries. The survey has four groups with the potential of answering up to 30 questions based on the response of the first and only mandatory question. The survey can take less than 10 minutes depending on how in depth of details a surveyee would like to provide. The overarching themes and ideas of the survey are the following:

Understanding User Needs and Satisfaction

The survey kicks off by categorizing respondents based on their use of IT: work/business, home/hobby, or both. This segmentation helps tailor the approach to better serve the needs of the surveyee.

For those in the work/business category, the survey delves into IT services satisfaction. Those not satisfied have options to suggest improvement, and it explores the idea of future growth prospects for Linux.

Home and Hobby Enthusiasts

For those using Linux at home or for hobbies, the survey asks about their preferences among openSUSE’s offerings like Tumbleweed, Leap, MicroOS, and SlowRoll, and why these distributions appeal to users and developers.

Contributions and Open Source Involvement

As an open-source project, it makes sense to inquire about contributions to open-source projects and to openSUSE. This helps to recognize not only the use of open-source software and projects, but how much users are contributing to it.

General Insights and Future Trends

This section asks all respondents about broader use of Linux beyond primary use cases. The aim is to help to understand the benefits respondents experience and the challenges they encounter. Respondents are encouraged to share their preferences for specific Linux distributions and their thoughts on emerging IT trends and technologies. The community members who came up with the survey are keen to understand if Linux is well-positioned to meet the market’s evolving needs with medium to long-term strategies.

The openSUSE’s Use Case Survey is not just a set of questions; it’s a collective effort to ensure that Linux continues to evolve to meet the diverse needs of its users and to make a brighter future for open-source technologies. Stay tuned for more updates and analysis from the results of the survey, which will run until Oct. 31.

After publishing Agama 3 a month ago, it is time for a new release. Among other things, this new version fixes several issues in the startup process, allows the use of a network proxy, adds (partial) support for IPv6, includes a few improvements in the web UI and features a new tool to extract Agama-related logs.

But as important as those changes, we did some internal work that will serve as the base for other features, like proper i18n support and more flexible storage management. Some of those features will land sooner than later, but meanwhile, let's focus on what Agama 4 brings.

Agama startup issues​

Our QA team is testing Agama through several scenarios and in different architectures. Those tests helped identify some issues with Agama's startup process: D-Bus activation errors, time-outs and even potential crashes.

Working closely with QA, we have identified and (hopefully) fixed most of those problems. In a nutshell, you should not see the "Cannot connect to D-Bus" error message anymore. And if that's not the case, please open a bug report attaching the logs 😉

If you are interested in the technical details, there is a bunch of fixes you might want to check: #729, #732, #747, #749, #753.

Installing through a proxy​

Many users and customers are used to deploy their systems with no direct access to Internet, just using a proxy. So it did not come as a surprise that it was one of the most requested features. Agama 4 includes support for specifying a proxy at boot time using the option proxy= when the installation requires to use a HTTP(S) or FTP source. The supported proxy URL format is: protocol://[user[:password]@]host[:port].

Of course, the proxy configuration is copied to the target system at the end of the installation.

The nitty-gritty details are available in #696 and #711.

IPv6 support in the automated installation​

In Agama 4 it is possible to specify the IPv6 configuration when using the automated installation. To accommodate this feature, we decided to introduce some changes in the profile definition. Here is an example that sets up IPv4 and IPv6 addresses.

  "network": {
    "connections": [
      {
        "id": "Ethernet network device 1",
        "method4": "manual",
        "method6": "manual",
        "addresses": [
          "192.168.122.100/24",
          "::ffff:c0a8:7ac7/64"
        ],
        "gateway4": "192.168.122.1",
        "gateway6": "::ffff:c0a8:7a01",
        "nameservers": [
          "192.168.122.1",
          "2001:4860:4860::8888"
        ]
      }
    ]
  }

Matching specific network devices​

In previous versions of Agama, it was not possible specify which interface to use for a given network connection when using the auto-installation mechanism: Agama just delegated on NetworkManager to decide. Now you can associate a network connection with an specific interface by using its name or more complex conditions. See #723 for further details.

  "network": {
    "connections": [
      {
        "id": "Ethernet network device 1",
        "method4": "auto",
        "interface": "enp1s0"
      }
    ]
  }

Agama specific storage settings​

Agama reads the information about the products offered for installation from a configuration (/etc/agama.yaml). Such a file contains a storage section which indicates the options (whether to use LVM, encryption, etc.) and the volumes (file systems) to create in the target system.

Since both Agama and YaST use the same mechanism to calculate the storage layout (known as the Guided Setup in YaST), the configuration for Agama was a direct translation of the YaST settings. But although the internal components and algorithms are the same, Agama's approach for tweaking their behavior is actually different in several aspects from YaST. For that reason, Agama now implements its own storage settings, making the product configuration more straightforward and less error-prone.

This change has a direct impact in Agama's auto-installation profiles.

storage: {
  bootDevice: "/dev/vda",
  lvm: true,
  encryptionPassword: "123456"
}

Interested in the implementation? Feel free to check: #721, #738 and #748.

Better logging support​

As the project evolves and more people try Agama, we need to make the debugging process easier. Recently, we introduced a new command that gathers all the information we need to debug the problems you face when using Agama (similar to the venerable save_y2logs for YaST). Typing agama logs store creates a tarball you can attach to your bug reports.

This first version (#757) is rather basic but we are already working on additional features to have a pleasant debugging process 🐛.

Polishing the web user interface​

Balsa Asanovic has become a regular contributor of Agama. For example, he implemented the "show password" feature (#750) and improved the error reporting in the iSCSI form (#699). But not only that, he is actively involved in our discussions in GitHub. Thank you, Balsa!

Moreover, a bug that prevented to download YaST2 logs through the web UI has been fixed (#746).

i18n support is coming to Agama​

If you follow the project closely enough, you might already know that we are working on i18n support. Many of the pieces are already in place but we did not make it for this version. However, we trust that our next major version will feature a translated web interface. Actually, our translators are already working on that.

Trying Agama 4​

The easiest way to try Agama is to download one of the two variants (ALP or openSUSE) of the Agama Live devel ISO. This image is built in the systemsmanagement:Agama:Devel OBS project and is updated each time we release a new version.

If you are interested in the bleeding edge, try the ISO in the systemsmanagement:Agama:Staging OBS project. It is built automatically and contains the code from Agama's Git repository, so it might get broken occassionally.

What to expect​

We expect to be able to ship a translated Agama interface and better handling of the storage settings in the next release. But, meanwhile, we are already working on other important features like software patterns selection or support for the SUSE Customer Center.

Of course, we appreciate opinions and feedback. As usual, feel free to contact the YaST team at the YaST Development mailing list, our #yast channel at Libera.chat or even the Agama project at GitHub.

Stay tuned!