I wanted to ask how far you have come to being able to update a model on Shapediver. Always having to upload a new model, copying the settings, etc. is just tedious and must also mean you are using way more resources than necessary. Out of the 136 models we currently have in our Library on Shapediver, there are maybe 4 or 5 distinct ones. Everything else is just small updates and iterations where I would have happily just replaced the Grasshopper file of the model, but left everything else - including the ticket code - untouched.
I would estimate that I have spent at least 2 hours of my life, just copying the shapediver codes back and forth, even though with a replace feature I would not have needed any of that.
You promised this feature would come a long time ago. How far away are we from this?
Okay, I guess that’s not coming any more. I also remember you used to be able to upload and configure directly from Grasshopper, but I guess that’s also not coming back!?
Regarding updating models: some version of this feature is definitely coming, and the reason it is taking a while is that we are gathering use cases to make sure this feature solves several problems rather than a specific one. There are several pain points here:
Model duplication in the library
Having to import settings from a previous model manually
Having to replace the ticket in models embedded outside of the platform
Potentially many more…
Additionally, a bigger issue that we have been wanting to tackle is proper versioning of models, which goes beyond the simple re-upload feature which we used to have in the old platform. Imagine being able to roll back to previous versions in case of issues, or even use a model for dev, staging and production environments while you develop your web application
The issue in your case is clear, and I thank you for your feedback. As we make progress, we will make sure to gather feedback from the community and develop this feature in the best way possible.
Regarding the configuration and upload of models directly from Grasshopper: this was never part of the ShapeDiver product but a version of this feature was present in a beta testing phase that you were a part of. The feedback we got along with a need to rebuild part of our infrastructure mean that not only has this planned feature been delayed, it will also likely take a very different form. If you have followed the latest developments we did regarding desktop clients, you can imagine how we plan to use this new approach to allow a live link between building a Grasshopper definition and seeing the results in the platform, configuring the model and uploading it.
The common thread here is that we do not think in terms of features but in terms of users and the problems they have, and the solutions we come up with often end up being quite different from what one envisions as a quick fix to a specific issue. Keep in touch with our latest developments by checking regularly the release notes in our help center, and I will make sure to post updates in this thread once progress is made about any of the two topics you mentioned.
Thanks @mathieu1 for the very honest and detailed explanation. It all sounds very sensible and I look forward to using it in the future.
As for versioning I totally agree that this would be almost necessary once you can update a model. I think there could always be an ticket code for production (always uses latest model), development and then you create a specific ticket code for each version, so its easy to roll back.
The help center is great, but I think there should be more examples. Examples make it so much easier to learn and understand things than just documenting with no context. By the way, I “taught” GPT4 a lot of the new v3 viewer API by giving it a lot of examples and documentation and now its pretty badass at writing typescript for the viewer API. But it obviously can’t produce good results for things that do not have examples.
Any updates on this @shapediver ? Working with ShapeDiver is a constant ping-pong of iterations (appearance tuning, client’s review, performance etc.). I am forced to set all settings over and over again on each upload and it is painful and I have to point out that we are talking about paid service! This is why I would expect this feature to be available. beacuse script iterations are somehow very common among computational design workflow.
BUG: Re-uploading the same script with a different name of the file leads to geometry problems. Some results probably stay cached in the memory? I have no clue, but I always have to delete previous files to get rid of that nasty problem. Below you could see text overlapping a previous version (even tho the gh filename is different!!!)
You are not forced to set all settings over and over again on each upload. While it is true that proper versioning is not yet available, you can use the “Import parameter settings” and “Import viewer settings” functionalities of the model edit page to copy all settings (parameter values and viewer configuration) from an old model to a new one. This is still something you have to do at each upload, but much more convenient than manually setting each setting one by one. You can read more about this feature here: Import model settings
About the bug you are reporting: please save a file that shows the issue and send the link to me. It might be that an unknown bug is happening, but I think this is just a known issue that we are in the process of fixing: sometimes, when you open a new model on the platform, the geometry of the previously opened model gets mixed in the viewer with the new one. In this case, this is not a bug in your model, just a small display issue which can be fixed by simply reloading the page. I will let you know when this is fixed on our side.
I agree that having to copy over the settings every time and uploading any update to a separate model seems silly to me too. We have now something like 230 models on Shapediver even though its really just 1 that has 230 iterations.
The fact that there is no way to simply update a model and keep all the API keys the same means we have to keep a separate Google Sheet where I always have to copy the latest API keys so the devs can then copy it into the code again.
If there was a way that I could update a model and didn’t have to do anything except upload a new .gh file, that would be really helpful.
