I have been working with Solr for a while, mainly from the .NET world and I basically love it. I use SolrNet which I think it is a very mature and stable library. I was asked today if I have ever used SolrExpress and if I recommend it over SolrNet.
The short answer is no, I have not used it. Therefore I can’t give a facts based recommendation, but looking over the source code of both libraries it is my opinion that SolrNet is still more complete. So I still believe SolrNet to be a more sensible choice.
It is worth mentioning that is a biased point of view, as I have used SolrNet multiple times and it really has made my life a lot easier.
Having said that, besides using it several times, I have authored a few things around Solr and SolrNet and used it extensively. It works fine and I know it pretty well. It basically gets the job done, it is pretty mature and almost complete (pending SolrCloud and a few minor things like a breaking change on collation).
Some of the things I created
I created a Solr training for Pluralsight

https://www.pluralsight.com/courses/enterprise-search-using-apache-solr
Getting Started with Enterprise Search Using Apache Solr …
www.pluralsight.com
Search is one of the most misunderstood functionalities in the IT industry. Apache Solr brings high quality Enterprise Search to the masses.
And a SolrNet training for Pluralsight
https://www.pluralsight.com/courses/implementing-search-dotnet-applications

I wrote a book for a company called SyncFusion for their Succinctly Series for Solr and SolrNet
https://www.syncfusion.com/resources/techportal/ebooks/apachesolr

I’ve also done internal trainings, presentations and webcasts on Solr + SolrNet
http://www.meetup.com/Atlanta-Net-User-Group/events/222161640/

Learn How to Add Search to .NET with Solr & SolrNet …
www.meetup.com
Search is a functionality that most people take for granted while at the same time it is deeply misunderstood and usually poorly implemented. .NET
SolrNet does not have yet support for SolrCloud in the main repository, but there is one fork that already uses it but our current project does not use forks, only the main repository. If that is not a blocker for your customer, go ahead or like in our case, just use a load balancer for querying and a call to zookeeper api to get leader for indexing.
Hope this helps.