Skip to Content

Contributors

pandoc-*.deb cleaning in OCA repositories

Due to a mistake on the tool to generate the READMES (fixed in [1]), some very big files ~30 MB each~ (with extension .deb or .dmg for Mac users) have been added both by the merge bot or by users doing a module migration.

This makes that on branch/repository clonation, or any free pull operation on any local git repository copy (of any branch) wastes a lot of bandwidth and resources, and is not ecological friendly.

Due to this, and although not ideal, we plan to force push the affected repositories (note that this has been only in 17.0 branches), following [3] technique recommended by Nils.

The drawback of doing this is that your next `git pull` operations on that branches will fail, saying about unrelated commit histories or creating a merge commit with possible conflicts.

The solution for avoiding it is to do the commands:

git fetch origin 17.0
git reset --hard origin/17.0

(being origin the OCA remote). In fact, this is the recommended way to do it in automated pulling systems.

If you don't have any strong counter-arguments, I will perform it at the end of the week. I will announce here the affected repositories after the operation.

Regards.


by Pedro M. Baeza - 06:11 - 29 Jul 2024

Follow-Ups

  • Re: pandoc-*.deb cleaning in OCA repositories

    pre-commit has a check to avoid merging big files


    IMHO we should enable/configure it in order to avoid a similar discussion about in the future

    by Moisés López Calderón - 04:11 - 7 Aug 2024
  • Re: pandoc-*.deb cleaning in OCA repositories
    OK, I desist. I think we are being a bit selfish, but OK, I have a lot of things to fight for.

    Regards.

    by Pedro M. Baeza - 03:41 - 7 Aug 2024
  • Re: pandoc-*.deb cleaning in OCA repositories
    Exactly. We should be able to come up with an additional git filter to keep the files out of the future series branches.

    Please consider leaving the branches as they are. It will break all setups that have forks of OCA modules included as submodules. On many Odoo.sh projects, this is the preferred way of working and typically these forks contain forward merges of pending PRs. The proposed rebase of the series branches would require a reconstruction from scratch of the corresponding branches on these forks.

    Cheers,
    Stefan

    On 07-08-2024 13:38, Jairo Llopis wrote:
    El lun,  5 de ago de 2024 a las 12:42:29 P. M., Adam Heinz <notifications@odoo-community.org> escribió:
    Add a note to the migration documentation to delete the large files commit?

    Yeah, probably you can just exclude .deb files in the format-patch or am commands.

    _______________________________________________
    Mailing-List: https://odoo-community.org/groups/contributors-15
    Post to: mailto:contributors@odoo-community.org
    Unsubscribe: https://odoo-community.org/groups?unsubscribe



    -- 
    Opener B.V. - Business solutions driven by open source collaboration
    
    Stefan Rijnhart - Consultant/developer
    
    mail: stefan@opener.amsterdam
    tel: +31 (0) 6 1447 8606
    web: https://opener.amsterdam

    by "Stefan Rijnhart" <stefan@opener.amsterdam> - 03:15 - 7 Aug 2024
  • Re: pandoc-*.deb cleaning in OCA repositories
    El lun,  5 de ago de 2024 a las 12:42:29 P. M., Adam Heinz <notifications@odoo-community.org> escribió:
    Add a note to the migration documentation to delete the large files commit?

    Yeah, probably you can just exclude .deb files in the format-patch or am commands.

    by Jairo Llopis - 01:36 - 7 Aug 2024
  • Re: pandoc-*.deb cleaning in OCA repositories
    Add a note to the migration documentation to delete the large files commit? There is already a process to merge translation commits together during migration and this seems like a similar amount of difficulty.

    On Mon, Aug 5, 2024 at 8:32 AM Enric Tobella Alomar <notifications@odoo-community.org> wrote:
    The problem is that this files and changes will be imported on future migration PRs on future branches (some files are inside the module folder). This is a decision that will impact the future of all OCA users. So this is something that will not disappear with time. Letting it go because we think it will disappear is a mistake.

    Change nothing is the easiest choice but we all know that this will not disappear and is a bad choice. We also are discussing this, but I think that there is not so many repositories affected.

    On Mon, 5 Aug 2024, 12:17 Jairo Llopis, <notifications@odoo-community.org> wrote:
    El vie,  2 de ago de 2024 a las 12:52:18 P. M., Enric Tobella Alomar <notifications@odoo-community.org> escribió:
    I think the benefits are bigger than the cons

    I think the opposite.

    Benefits:
    1. Save disk space in case you're either not cloning with --depth 1, or you're going to do a git gc after all the manual work that will mean updating all your local clones.
    2. Save bandwith space under similar conditions.
    Potential problems:
    1. Break your lock files.
    2. Need to rebase almost every open PR against those branches.
    3. Inclusion of unwanted commits (that include those wonderful 30MB files) in more PRs.
    4. Disturbing many developers.
    5. Making oca-port unable to detect relationship between commits and PRs.
    6. Manual work involving the repo admins and almost all OCA developers.
    7. Things I don't imagine right now.
    Definitely not worth it. One day we'll stop using those branches (hopefully in about 3-5 years) and we won't care about those few extra MBs.

    _______________________________________________
    Mailing-List: https://odoo-community.org/groups/contributors-15
    Post to: mailto:contributors@odoo-community.org
    Unsubscribe: https://odoo-community.org/groups?unsubscribe

    _______________________________________________
    Mailing-List: https://odoo-community.org/groups/contributors-15
    Post to: mailto:contributors@odoo-community.org
    Unsubscribe: https://odoo-community.org/groups?unsubscribe


    by Adam Heinz - 02:41 - 5 Aug 2024