Axel Huebl

I did not try to modify the cache yet but are now at a point where even selecting a few dozen classes (out of many 100; number of doxygen xml...

digging a bit deeper, the underlying problem seems to be that single entries in the cache can easily get up to MBytes in size. The main consumers seem to be...

the interesting thing is that the `.xml` files that doxygen creates, with all their well-known formatting overhead, only sum up to `84 MByte` for the project I am reporting. I...

I think the main issue is the [memory footprint](http://stackoverflow.com/questions/11127529/memoryerror-with-minidom-in-python) of the xml parser `minidom`, which's [parsed result object](https://github.com/michaeljones/breathe/blob/d3eae7fac4d2ead062070fd149ec8bf839f74ed5/breathe/parser/index.py#L47-L60) is cached. I think we have two ways to speed things up...

Since all the reports on `minidom` are discouraging (huge overheads in general, supposedly also quite some memleaks) from what I found on stackoverflow, I started investigating for B). I tried...

@michaeljones although I don't have the time to work on this deeper (sorry!), I think this issue should become some priority for the project since it needs coordination and is...

@svenevs since you are working on `exhale`, did you see the same performance issues with large projects with breathe? :) My try to get rid of `minidom` in breathe are...

Thanks for the detailed answer! > does the ENTIRE XML DOM get loaded into memory at once? the main problem is not that the entire XML dom gets loaded at...

Thx for giving it a try! Yes, I think the underlying problem is actually just `1.` since each and every `minidom` XML read is memory hungry/leaking. `doxygenindex` just happens to...

@svenevs @michaeljones @SylvainCorlay did by coincidence any of you try to reduce the memory footprint of breathe? On RTD, I get regularly killed in builds due to the leaks in...