home.social

Ankur Sinha "FranciscoD"

View on fosstodon.org
  1. Hacked up a new script to update multiple repositories (or just the one) non-destructively. Doesn't handle all use cases, of course, but useful for a lot of general cases. It stashes any currently uncommitted work, and updates multiple branches---which can be specified as arguments.

    github.com/sanjayankur31/100_d

  2. Upgraded another machine to @fedora 44. Upgrade went well, no issues. on this machine wouldn't launch. Looks like an issue and GPUs, since this one has an GPU. Workaround seems to be to play with an environment variable. This worked for me: `QTWEBENGINE_FORCE_USE_GBM=1 qutebrowser`

    References:

    github.com/qutebrowser/qutebro

  3. PSA for users ( based): we need to use this option to enable support in `~/.config/qutebrowser/autoconfig.yaml` (for example, if you're a @fedora package maintainer and don't want to have to keep re-logging into Fedora websites):

    ```
    qt.args:
    global:
    - auth-server-allowlist=*<domain>
    ```

    Used to be `auth-server-whitelist` but changed at some point.

    Reference: github.com/qutebrowser/qutebro

  4. Moved to a new indicator that works well:

    extensions.gnome.org/extension

    (The previous one I was using doesn't support the latest Gnome version)

  5. Building detailed neuronal models is hard--- helps, but there’s still a steep learning curve. We're investigating how can be used to create natural language interfaces to improve accessibility:

    - for NeuroML: learn from and query curated information sources
    - iterative LLM assisted code/model generation: modelling, validation, and simulation

    More details in the post:

    ankursinha.in/2026/01/30/build

  6. Note to self: use `pre-commit autoupdate` to update pre-commit repositories to their latest versions.

    -commit

  7. @dzwiedziu

    It does allow you to specify what project you want to add the new tasks to. But, please feel free to tweak it for your particular needs.

    I am aware of the python libraries. There's `tasklib` that the taskwarrior upstream maintains (and is used by other task related tools I use like ) :

    github.com/GothenburgBitFactor

    I thought these felt a bit overkill for this simple script, but they can certainly be used. Here I rely on plain parsing of the json export.

  8. How do you manage your task list? If you're not happy with your current setup, give a try---it's a really lean, simple yet powerful and extensible tool for managing tasks. It also includes in-built syncing features so that you can use across multiple devices.

    taskwarrior.org

    Recently, I needed to split tasks into smaller tasks. I came up with a quick script:

    ankursinha.in/2025/07/05/split

  9. Quick note: if you are using (even via ), we're seeing a bug with the new version 3.0.6 where it doesn't recognise the encryption password. Reported upstream here:

    gitlab.com/duplicity/duplicity

    So, do if you see this issue too, downgrade to 3.0.5. On @fedora , this would be `sudo dnf downgrade duplicity`.

  10. Reminder to self: listing/opening "recent files" in : `:oldfiles` or `:History` with fzf.vim

    stackoverflow.com/q/309723/375

  11. I can't say this enough: time and time again the has come through to look after people, some that i know and loads that i don't. I know there are issues that we see amd read about, but i am grateful to all the people that work tirelesslely all hours to help strangers that come in for help. So, . We are lucky to have you.

  12. A new release of is available. Please update to get the latest fixes and features.

    ```
    pip install --upgrade pyneuroml
    ```

  13. If you're seeing your simulator builds fail recently, it's because of a new release:

    github.com/nest/nest-simulator

    Workaround: fix `cython<3.1.0` for the moment (until new NEST releases with the fix are made)

  14. Generated a quick plot to see usage trends for Open Source Brain v2 (): an integrated platform for that indexes multiple model and data sources (, , , ) and provides compute resources on the in "workspaces". It also includes specialist applications: for working with data in the NeuroData Without Borders () format; -UI for biophysically detailed and a environment.

  15. Updated to the latest version in my repository for @fedora :

    copr.fedorainfracloud.org/copr

    TaskJuggler is an Open Source project management tool. It's full of great features---scheduling, resource allocation, gannt charts, reports and reporting, all of it. See the documentation here: taskjuggler.org/documentation.

    I use it to manage all my work projects, and it's a real boon. Give it a go!

  16. If you're using a based browser like me (@qutebrowser or ), 's website may cause issues because they "only support the last 3 releases" of most browsers. Luckily, they use the user agent to detect the browser. So, a workaround is to use something like this, with a new enough version for the version:

    Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) QtWebEngine/6.8.2 Chrome/132.0.0.0 Safari/537.36

  17. Note to self: a board is NOT a uni-directional pipeline!! It's OK for stuff to go from "WIP" back to "Ready" or "Backlog"

  18. is participating in again this year under @INCF . We're looking for people with some experience of to work on developing biophysically detailed computational models using and .

    Please spread the word, especially to students interested in modelling. We will help them learn the NeuroML ecosystem so they can use its standardised pipeline in their work.

    docs.neuroml.org/NeuroMLOrg/Ou

    CC

  19. supports all stages of the modelling life-cycle with a vast ecosystem of software tools: creating (, , , , , ), validating (pyNeuroML, , ), visualising (pyNeuroML, , -DB), simulating (, , , , , , ), model fitting/optimisation (, , NetPyNE), sharing and reusing of models (OSB, NeuroML-DB, .org). 7/x

  20. A number of software tools are available for construction and simulation of models: , , , , , , etc. These have their own features, styles, programming interfaces (APIs). This is great but it also means that researchers need to learn each of these individually to use them. It also means that tools and models developed for one don’t necessarily work for others and need to be manually converted. This is often a non-trivial task and limits model reuse. 3/x

  21. We are very happy to provide a consolidated update on the ecosystem in our @eLife paper, “The NeuroML ecosystem for standardized multi-scale modeling in neuroscience”: doi.org/10.7554/eLife.95135.3

    is a standard and software ecosystem for data-driven biophysically detailed endorsed by the @INCF and CoMBINE, and includes a large community of users and software developers.

    1/x

  22. How's Friday going folks? I've been working on standardising the Cell model from Zang et al 2018 to . Exported the morphology already, on to the converting the individual Ion channels next.

    You can use the converted bits from here: github.com/sanjayankur31/ZangE

  23. Very well said!

    Shared from birdsite:

    x.com/lorraine/status/17276515

    “I’m tired Lorraine of having to justify my existence by the worth I bring to this country”

    DrAmirKhanGP
    responds to the immigration and NHS claims made by Nigel Farage in last night’s episode of
    imacelebrity

  24. I use pushd/popd quite a bit to navigate directories. I've also been using quite a bit with and without . It's Alt + C mapping for the terminal is awesome. So I went ahead and created a new Alt + P mapping that does the same thing but uses instead of . Here's the tweak to my `~/.bashrc` if anyone else wants to do it too:

    github.com/sanjayankur31/100_d