Are frames-within-frames reliable, or are they best avoided at present for production code?

I'm using frames on notebook pages to good effect, including creating and destroying them at runtime as pages are added and removed. However this morning I identified a visual control which could be usefully put in its own frame with some extra methods and appear to have run into problems.

Despite the fact that the editing was done in 0.9.28, it won't reload the project, telling me "Unable to find the unit of component class...". 0.9.29 will load and compile, but I'm not sure that the embedded control is being referenced correctly.

Should I persevere, or is this still experimental and best avoided?

Mark Morgan Lloyd wrote

Are frames-within-frames reliable, or are they best avoided at present for production code?

During 0.9.29 we have fixed many bugs in the related code both in Lazarus and FPC. I specially tested that combination in our work application but this does not give you any guaranty that they work perfectly for your case. They still need more testing.

Best regards,
Paul Ishenin.

We have successfully used a 2-level nested frame setup - placed on a pagecontrol. It works like a charm... :)

Kind regards,
Torsten Bonde Christiansen.
EpiData Association.

Thanks Paul and Torsten.

Does that mean that I should be using a version of the FPC compiler more recent than the 2.4.0 release?

I'm very much in two minds over this. Noting Torsten's success, if I could get the nesting working reliably then it would be extremely useful, OTOH I'd quite like the code to compile with something that's not quite "bleeding edge".

Mark Morgan Lloyd wrote

Does that mean that I should be using a version of the FPC compiler more recent than the 2.4.0 release?

Yes, you need 2.4.2 release which is in the preparations step now. And you also need Lazarus 0.9.30 which is still in the bug fixing stage.

In other way you can get a situation when you need to edit lfm and lrs files manually.

Best regards,
Paul Ishenin.

Thanks Paul, noted. I think in practice that suggests that I might be better off not using the facility for the moment, at least until I have another of my periodic sessions checking FPC+Lazarus on all available platforms (and no doubt irritating people enormously when I ask about problems on minority processors :-)

Mark Morgan Lloyd wrote

... at least until I have another of my periodic sessions checking FPC+Lazarus on all available platforms

This feature is platform independent, so if it work for you on one platform on others it will work too. Only the fpc and lazarus versions should match.

Best regards,
Paul Ishenin.

Sorry, what I meant was that there might be some odd combination like SPARC/Linux that isn't worth getting Lazarus working on. At that point I'd prefer keeping my source in a state where it would compile for any machine that I'm running here.