Most of my projects these days are a maximum of 10 mins long when cut, so I don't feel the limitations of Kdenlive's simplistic structure too much. ie. Kdenlive lacks editable nestable sequences, multiple sequences within a project, has only one level of folders/bins, no ability to export sequences to interchange formats (EDL, OMF) for collaboration...
I've got a big, complex documentary project coming up and am unsure about my workflow. I know how I would approach this with FCP/AVID/Premiere but I don't want to go back to proprietary software.
I hope I can learn something from other users - how do you handle long, multi-scene projects in Kdenlive? What's the longest or most editorially complex project you've undertaken? When did you feel you were up against Kdenlive's limits, and how did you deal with that?

(Note that I'm mostly talking in terms of organisation - how do you maintain an overview of your footage, how do you maintain flexibility in your edits (non-destructive editing), how do you create/ maintain different versions of your edit for different purposes/lengths/formats?)

[EDIT: as zekthedeadcow pointed out, you can nest .kdenlive files in other .kdenlive files. The workflow isn't quite the same as in other NLEs but the end result is basically exactly the same.]

I am currently editing a 12 episode webseries... each episode is 5-15 minutes and the total project was 99 pages.

Sane naming conventions and good camera notes are a big help.

Kdenlive projects are nestable. I cut each episode as a separate project but the .kdenlive files can be loaded as clips to assemble it as a feature if i wanted.

I have also started developing a "production manual" to store note on how things are done for each step of production... from forms for casting and locations... to how process files for archival and distribution. It saves a lot of time because I no longer have to remember what my archival process is and what the command lines entries are I post each project on a production kanban... which is rather large... two 4x8 sheets of melamine, lables are white mixer marking tape (so it can be rearranged) ... so I know what the next step for each project is.

For a larger project I filmed roughly 1700 clips in nine days. Each day I checked the footage and renamed clips that I could possibly use to something meaningful that I could then find on the filesystem easily.

Then I set up the kdenlive project and put the .kdenlive file into a version management system (mercurial in my case). So I could try different cutting versions and restore the old files when needed. The advantage is that you do not need to copy and rename the .kdenlive file whenever you want to try something. Commit changes to your version management system if a logical unit of work is done.

During production I searched the clips on the file system and checked which one to import to the project by playing them with a video player (HD with h/w acceleration). Then I imported the needed clips into the project and created proxie files. Then I did the editing and finally removed any file that I did not need for cutting from the project in order to keep it small.

Intro and credits were seprate projects and have been rendered long before the main movie was done. If you choose the same render profile you can easily concatenate the rendered elements losslessly with ffmpeg. That saved rendering time, as all versions have the same intro but the credits were only for a downloadable version.

So, finally I fully agree to zekthedeadcow: Good planing of the project and good clip naming conventions are very helpful. You can divide the project into pieces and either use their .kdenlive files as clips or render and concatenate the parts. Using a version management system for the .kdenlive files helps to keep the working directory tidy without .kdenlive files for each day or different cut.

Thank you very much for the tips!
I have been playing around with nesting projects today, so if I can work out the right sized 'modules' and naming conventions I think it will help a great deal.
I also spent a while digging into my kdenlive projects in a text editor (fascinating, just fascinating ). I didn't necessarily achieve anything by doing so, but I feel like I understand kdenlive a lot more now!

One nice thing I discovered in my experiments is that the combination of nesting and proxies is actually a good dirty workaround to render sections of the timeline - when I've got animation PNG sequences or titles overlaid on video clips with effects applied, there's no hope of things playing back without stuttering, even when using proxies. BUT if you create your complicated section in one .kdenlive file (without using proxies) and then nest it in another .kdenlive file (this time using proxies) then the newly created proxy gives you nice, smooth, uncomplicated playback, and you can quickly and easily update proxies if you change the nested file.

And good idea using version control, capslock - I'm only just getting started with git (my first project is open sourcing an animation) so I'm looking forward to trying it with editing projects as well. I've also been doing a lot of concatenating with ffmpeg, it's a lifesaver!

I haven't used a kanban for production planning before, maybe it's worth having a play with a software version to see if it works for me. Anyone know if Libreboard is any good?