The website does not emphasize enough that Stratosphere has its own MapReduce runtime and does not use Hadoop's MapReduce.

At the beginning I also through that Stratosphere uses Hadoop.

Some statements like

"It combines the strengths of MapReduce/Hadoop with powerful programming abstractions in Java" on the first page.

and

"Stratosphere for Hadoop 1/Hadoop 2" in the download section are very misleading.

In my opinion in the download section we should completely leave out the term "Hadoop" and use "HDFS" instead.

Maybe we can prevent questions like the most recent one:
https://groups.google.com/forum/#!topic/stratosphere-dev/-WSxxtsdCSo

Please mention that the environments and data sets are in eu.stratosphere.api.java.

The snapshot quickstarts are only in sonatype, not in maven central. To directly use the 0.5-SNAPSHOT quickstarts, one needs to add the sonatype repo to the known repositories.

The website does currently not describe how to do so.

I cannot make a pull request for this section, because I am on mobile internet and cannot afford to clone the stratosphere.github.io repository. I have pasted my text below. Sorry Ufuk, for causing additional work.

Introduction

Analysis programs in Stratosphere's are regular Java Programs that implement transformations on data sets (e.g., filtering, , mapping, joining, grouping). The data sets are initially created from certain sources (e.g., by reading files, or from collections). The results are returned by sinks, which may for example write the data to (distributed) files, or print it to the command line. The sections on the program skeleton and transformations show the general template of a program and describe the available transformations.

Stratosphere programs can run in a variety of contexts, for example locally as standalone programs, locally embedded in other programs, or on clusters of many machines (see [program skeleton] how to define different environments). All programs are executed lazily: When the program is run and the transformation method on the data set is invoked, it creates a specific transformation operation. That transformation operation is only executed once program execution is triggered on the environment. Whether the program is executed locally or on a cluster depends on the environment of the program.

In contrast to the Stratospheres Record API, the Java API is strongly typed: All data sets and transformations accept typed elements rather than generic records. This allows to catch typing errors very early and supports safe refactoring of programs.

Can someone please confirm that the debian package is updated for release-0.5?

In: http://stratosphere.eu/docs/0.5/programming_guides/java.html

• remove "public" from MyCoGrouper or insert "static" and wrap plan construction code in main() method
• "reduceGroup" should be "with"

http://stratosphere.eu/quickstart/java.html
the curl ... .sh url is wrong

I think more people will have noticed the following issue with the navbar on the left:

It still kinda works, but I find it inconvenient. The problem did also occur before, but wasn't so bad. I suspect that now almost everybody working from a Laptop will experience it... especially since people will likely look into the Java API.

The questions are the following:

• Did you experience the problem? How inconvenient is this for you?
• Should we move the subnavigation the the top of each article (as seen here)? This would also allow to have more fine-grained sub bullets when needed (e.g. data transformations => map, reduce, etc.) But we would loose the option to easily jump around in an article as you would have to go back to the top every time.

Just got this response from GitHub after pushing a small update:

The page build completed successfully, but returned the following warning:

GitHub Pages recently underwent some improvements (https://github.com/blog/1715-faster-more-awesome-github-pages) to make your site faster and more awesome, but we've noticed that stratosphere.eu isn't properly configured to take advantage of these new features. While your site will continue to work just fine, updating your domain's configuration offers some additional speed and performance benefits. Instructions on updating your site's IP address can be found at https://help.github.com/articles/setting-up-a-custom-domain-with-github-pages#step-2-configure-dns-records, and of course, you can always get in touch with a human at [email protected]. For the more technical minded folks who want to skip the help docs: your site's DNS records are pointed to a deprecated IP address.

For information on troubleshooting Jekyll see:

https://help.github.com/articles/using-jekyll-with-pages#troubleshooting

If you have any questions please contact us at https://github.com/contact.

docs/0.5/programming_guides/hadoop_compatability.html are not finished yet.

This message from our mailing list, posted by @fhueske might be a good skeleton:

Similar to Spark, Stratosphere is a complete data processing system, i.e., it has a programming API, a program compiler (optimizer), and an own execution runtime.
It is also an alternative for Hadoop MapReduce and in several design points quite similar to Spark:

• Programs are executed as DAGs
• Higher-level programming primitives (compared to Hadoop MR)
• APIs in Scala and Java
• Reads data from external data stores (has no own data storage), e.g., HDFS, S3, RDBMS, ...

However, Stratosphere is also different in some aspects:

• Database-inspired processing using pipelining, gradually going to disk if memory is not sufficient (Hybridhash Joins, external sorts)
• Sophisticated cost-based optimizer choosing execution strategies (broadcasting vs. partitioning, sort vs. hash joins, ...)
• Implemented in Java (in contrast to Spark which uses Scala)
• No intermediate result materialization in memory (this is on the roadmap)

Stratosphere and Spark can be rather seen as alternatives.
We do not build on any of Sparks components as we have our own programming API and execution engine.

docs/0.5/program_execution/cluster_execution.html is not finished yet

I think you don't have to specify the "run" command anymore:

http://stratosphere.eu/docs/0.5/program_execution/cli_client.html

• Highlight Java API section "Powerful Programming Interfaces"
• Address #34
• Check for correctness

The Delta Iteration documentation in the Scala API should be updated.

It only says:
"This is tad bit prototypical right now. Please contact us through one of the channels here if you are interested in working with it."

We should at least remove the link to the contact page and add links to the general iteration documentation and the Delta Iteration Scala Example.

The example section needs to be updated for the new Java API and refactored Java examples.

We need to update the example documentation on the website for this.
We should also link from the API documentation to examples that show the API features in action.

I can help with the required Jekyll code, but I have no clue how to nicely integrate it.

$ bin/stratosphere run
--jarfile ./examples/stratosphere-java-examples-0.5-WordCount.jar
--arguments 1 file://pwd/hamlet.txt file://pwd/wordcount-result.txt

should be

$ bin/stratosphere run
--jarfile ./examples/stratosphere-java-examples-0.5-WordCount.jar
file://pwd/hamlet.txt file://pwd/wordcount-result.txt

?

The new Java has put the sub navigation on top of the page. I think that other pages should also consider that.

The following links in the page are showing a 404 error. Please check.
http://stratosphere.eu/docs/0.4/program_execution/local_executer.html
http://stratosphere.eu/docs/0.4/program_execution/remote_executer.html