I can't find anything related to preview on my Arch Linux default Arch

kcm -> layouts -> add layout -> preview

I believe that port away from boost will fix the problem

I never had noticed any problems when compiling the code. Could it be that the CI system changed (e.g. newer compiler?) I would find it sad to lose this very useful feature (which might in fact not be exposed enough)

In D10118#196581, @graesslin wrote:

I never had noticed any problems when compiling the code. Could it be that the CI system changed (e.g. newer compiler?) I would find it sad to lose this very useful feature (which might in fact not be exposed enough)

This has been issue from forever.. see BUG: 362946

And for @hein and @broulik e.g this problem is so severe that it causes them to not able to compile plasma-desktop at all.

This locks up my PC several times a week. I lose work regularly because of it, and often can't finish compiling plasma-desktop successfully. I'm very +1 for killing it.

In D10118#196532, @anthonyfieroni wrote:

I believe that port away from boost will fix the problem

IIRC the problem isn't with boost at large, it's that the code uses insanely deep recursive templates.

In D10118#196820, @sitter wrote:

In D10118#196532, @anthonyfieroni wrote:

I believe that port away from boost will fix the problem

IIRC the problem isn't with boost at large, it's that the code uses insanely deep recursive templates.

Yeah:

# # the default maximum template expansion depth (256) is not enough
# set_property(SOURCE preview/geometry_parser.cpp APPEND_STRING PROPERTY COMPILE_FLAGS " -ftemplate-depth=512")

it is not clear if this code is functional at this moment or not

I'm -1 for disabling if the change is based on a false premise that it doesn't work.

Is it something you've already lost several hours into trying to fix? If so this proposal has a lot more merit.

If it is to be disabled, commenting something out when it's already in an if() block is not a good approach.
Also any hacks always always need a code comment.

FWIW: ktouch has it's own layout parser / preview code. It does update between UK and french and move the preview with the letters about, but switching to arabic returns an error, whereas the kcm preview works fine. Probably would need layer support too.

In D10118#196823, @davidedmundson wrote:

it is not clear if this code is functional at this moment or not

I'm -1 for disabling if the change is based on a false premise that it doesn't work.

No, this change is not proposed because I was not sure if is functional or not, but due to reason that it breaks my development workflow and probably for others as well..

Urgh, not sure why phab decided to kick all subscribers, sorry...

In D10118#196841, @sitter wrote:

eek. please use https://cmake.org/cmake/help/v3.0/command/option.html , and maybe add_feature_info (see https://cmake.org/cmake/help/v3.0/module/FeatureSummary.html) instead of the message

Line 5 does it already.

Where are we with this? The issue is really, really annoying and I imagine hits a lot of people trying to submit patches to plasma-desktop.

That said, I agree with @davidedmundson that we should actually fix the issue too. +1 on this for now though.

Here's an alternative approach that fixes the issue for me without needing to disable the feature: D17930