Layers and classifications serve a similar purpose, with layers being the older CAD-era paradigm and classification the newer, standard BIM way. It is somewhat of a nuisance that we still need to fill in and update both. For many long-term Archicad users, this seems like unnecessary double work and leads to confusion and neglect towards classifications, which means poor quality of models and extra work for BIM managers.
I propose to replace layers completely with classifications in Archicad. View management could be done similar to Layer palette by checking and unchecking branches of the classification(s) that are in use, and saved into Classification combinations (like layer combinations).
For compatibility with CAD import and export, some way of automated layer mapping would be required. Eg when importing from a format containing layers, Archicad could create a new (non-hierarchical) classification based on the layers of the original file, which would then enable controlling the visibility in the same way as described above. DWG translators could automatically export classification as DWG layers, with optional custom layer mappings (like now with Doors, Windows etc).
For external drawings linked with the Drawing tool, nothing needs to be done - the embedded layers can remain as they are now (this is our preferred method of incorporating DWGs anyway).
Replacing layer with classification is shortsighted. GS should instead move to a criteria based approach to control of element visibility, editability, intersectionality - as brought forward here. Then layers can be kept as is for compatibility with applications and workflows.
A lot of users still use layers for many reasons, you are free not to use the layers by deleting all of them and keeping only the Archicad layer. I had the opportunity to work with Revit but I stayed on Archicad because of the layers issue.
For example I use on my project two layers for furniture, lamps, etc, one for custom ones that have to be done in the project and others for furniture bought by client. Both are classified as furniture. When I send it to contractors a simply hide layers without losing the information.
Although I agree it's kind of old-autocad workflow, but it represents flexibility I don't want to lose.
I think I should clarify that I don't mean to enforce Archicad 2.0 classification only - it is quite possible and even useful to use a more specific and customized classification of your own, just the same way as companies and countries have had different layer standards in use. For instance, in Finland we have a national standard called Talo 2000 ("House 2000"), which is widely in use (with slight variations) in many offices for layer naming AND for classification, too. It has categories/layers for furniture, lamps, etc. So you could easily do the same kind of granular control for element visibility as is now done with layers. Using classification for this would be even better because of the tree-like structure, meaning that it is possible to have subcategories (branches), which makes the organisation more logical. Classes definitely don't need to correspond one-to-one with the default Archicad 2.0 classification (which is based on Ifc classes). Ifc type mapping can be then set up according to this custom classification. In fact we're not currently using Archicad 2.0 classification at all.
PS. Revit doesn't allow creating custom classification systems at all unlike Archicad and I agree we don't want such unflexibility in Archicad. With this proposed change, Archicad would have yet another area where it is on par, but more flexible and advanced than its main competitor.
Yes, there should still be a default classification and ifc type mapping for it coming with the generic template, and localised versions could have ones based on the national standard if one exists. These can then serve as a base for customized versions if such are wanted. Admittedly, setting the ifc type mapping from scratch can be a bit intimidating, since the the IFC specification is still a bit of mystery to many of us.
I don't disagree with the sentiment of the wish, but accommodating the legacy data into a fundamental rewrite of AC isn't going to happen any time soon.
To have classifications as the core display attribute you need to move to pure 3D modelling with the ability to section at any point for plans, elevations & sections and all annotation & 2D graphics would be applied by drawing, not the current hybrid arrangement AC has for plans.
If Nemetschek want to create a new NextGen AC with the ability to import legacy AC I'm listening, but for now I don't think it will happen. Let's face it they haven't even finished macOS Dark Mode compatibility in AC after three years or even Instancing after 20 years.
With the realities of software development, you're probably right. But one can always wish!
Regarding 2D graphics and annotations, I suppose these could be also included in the Archicad classification that would replace layers. IFC actually supports 2D data also (IfcAnnotation), as previously mentioned here. That part has so far just been omitted in Archicad (and Revit too I guess).
I have suggested a new layering system that allows for tagging. I see this with photos in Adobe Bridge, where you can use multiple tags for single elements and thus filter or group elements. But the tagging system would have a hierarchy too (may only necessary for export); 1. Interiors, 2. Walls, 3. Trim, 4. Base. Then this becomes a selectable attribute as well as a way to organize and filter like folders. And of course the tags are completely customizable and become part of the element's attributes. This is not fully thought out yet.
Tags could be a nice additional way to organize and control the visibility of stuff. But we still also need proper (IFC) classification to ensure the exported IFC models can be reliably machine-read and used for further analysis, calculations, facility management and archival. As I understand, these tools rely on the classification to be able to recognize building parts, otherwise they just see unorganized geometry and data. Tagging stuff, possibly with multiple arbitrary tags wouldn't work for automating the conversion to IFC classes, so we would still need to do classifying too. The less extra work this requires from the user, the better.