> On 22. Jul 2026, at 07:35, Gregory Casamento <[email protected]> wrote:
> 
> As our team grows, the binary nature of .gorm files is becoming untenable. 
> Because Git cannot merge binary files, concurrent changes result in merge 
> conflicts where "the last person to merge wins."

> While we can easily read these formats, writing to them will always be 
> imperfect. Therefore, I propose we design a custom, project-controlled 
> serialization format that meets three criteria:
>     1. Compositional Structure: Nested objects should reflect the actual UI 
> hierarchy (e.g., nesting an NSView inside an NSWindow tag).
>     2. Human-Readable & Editable: This will allow clean version control and 
> manual merging.
>     3. Project-Controlled: We dictate the schema, protecting us from external 
> breaking changes.

Hi,

This might be a bit out of left field, but is it perhaps feasible to
instead have a Gorm merge tool, keeping the format? I.e. you give it a
base/ours/theirs and it will merge the files with its specific
structural knowledge about Gorm files (maybe even built into Gorm
itself). Git already doesn’t have to use its built-in text merge, you can
use whatever tool you want.

I feel like generic text merge on a complex data format like UI layout
even if the format is theoretically suitable for it is always going to
be error-prone, especially if you aren’t an expert on reading the format
(and I feel like a merge tool would be more the NeXTSTEP approach :^) ).

Cheers,
Katalin

Attachment: signature.asc
Description: Message signed with OpenPGP

  • P... [email protected]
    • ... David Wetzel via Discussion list for the GNUstep programming environment
      • ... Daniel Santos
      • ... Riccardo Mottola
        • ... R Frith-Macdonald
          • ... Gregory Casamento
            • ... R Frith-Macdonald
              • ... Gregory Casamento
            • ... Katalin Rebhan
        • ... David Wetzel via Discussion list for the GNUstep programming environment
    • ... H. Nikolaus Schaller
    • ... Graham Lee
    • ... Riccardo Mottola

Reply via email to