I don't know many things about Spark, so I can't make a full comparison. I will try bellow to enumerate some advantages of Pippo:
- modular [1] (spark uses a giant static singleton architecture)
- support (via modules) for many Template Engines [2] (Freemarker, Jade, Groovy, Trimou, Pebble)
- support (via modules) for many Servers [3] (Jetty, Undertow, TJWS); spark depends on Jetty
- directly serving of static resources [4], support for WebJar
- content type negotiation
- custom session [5] (support for cookie based
implementation)
- support (via modules) for many ContentTypeEngines [6] (Plain text, Json - Gson/Fastjson, Xml - Jaxb/Xstream, Yaml - SnakeYaml)
- i18n [7] (in java or directly in template)
- support (via modules) for many IoC [8] (Guice, Spring)
- support metrics [9] (via modules - Ganglia, Graphite, InfluxDB, Librato)
- custom (template) error pages
- regex routes
Unfortunately the actual documentation of Pippo it's a little bit behind of the implementation (from documentation are missing some nice concepts: named routes, finally filters, custom route context, ...) and from this reason I cannot give you more useful links but I will fix this aspect asap.
Another difference in Pippo to other (micro) java web framework (spark, ninja) is that a RouteHandler it's a real endpoint (the handle method return void). You can finalize/commit the response using send, render, redirect. Everything is transparent.
[1] http://www.pippo.ro/doc/modularity.html
[2] http://www.pippo.ro/doc/templates.html
[3] http://www.pippo.ro/doc/server.html
[4] http://www.pippo.ro/doc/static-files.html
[5] http://www.pippo.ro/doc/session.html
[6] http://www.pippo.ro/doc/content-types.html
[7] http://www.pippo.ro/doc/internationalization.html
[8] http://www.pippo.ro/mod/spring.html
[9] http://www.pippo.ro/mod/metrics.html