[
https://issues.apache.org/jira/browse/BIGTOP-1680?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14326396#comment-14326396
]
RJ Nowling commented on BIGTOP-1680:
------------------------------------
Is there a clear recommendation for how users should handle application
dependencies so they conform with what BigTop installs?
This is an issue we faced at Red Hat for internally deploying applications. We
ended up creating local maven/ivy repositories when installing RPMs and
recommended that build systems resolve against the local repositories. I'm not
a fan of this solution as it requires the user to install the RPMs to build and
development applications -- also goes against Maven "philosophy."
If I'm correct, BigTop publishes its own JARs through Maven or similar?. Maybe
we should prefer to use these over the JARs projects publish themselves? This
way, when a user builds an application against version BigTop 0.8 or 0.9 JARs,
they will build against the same JARs deployed on the cluster?
> [BigPetStore] BPS-Spark Unit tests for transaction generation
> -------------------------------------------------------------
>
> Key: BIGTOP-1680
> URL: https://issues.apache.org/jira/browse/BIGTOP-1680
> Project: Bigtop
> Issue Type: Test
> Components: blueprints
> Affects Versions: backlog
> Reporter: jay vyas
> Assignee: RJ Nowling
> Fix For: backlog
>
>
> The following code could be split up into a method, i think, and would be
> greatly served by a companion unit test.
> Its the reference impl of the data generator on spark: i.e. this is where we
> magically generate all the customer transactions .
> {noformat}
> 155 val transactionRDD = customerRDD.mapPartitionsWithIndex{
> 156 (index, custIter) =>
> 157 // Create a new RNG
> 158 val seedFactory = new SeedFactory(nextSeed ^ index)
> 159 val transactionIter = custIter.map{
> 160 customer =>
> 161 val products = productBC.value
> 162 //Create a new purchasing profile.
> 163 val profileGen = new PurchasingProfileGenerator(products,
> seedFactory)
> 164 val profile = profileGen.generate()
> 165 val transGen = new TransactionGenerator(customer, profile,
> storesBC.value, products, seedFact ory)
> 166 var transactions : List[Transaction] = List()
> 167 var transaction = transGen.generate()
> 168
> 169 //Create a list of this customer's transactions for the time
> period
> 170 while(transaction.getDateTime() < simulationLength) {
> 171 transactions = transaction :: transactions
> 172 transaction = transGen.generate()
> 173 }
> 174 //The final result, we return the list of transactions produced
> above.
> 175 transactions
> 176 }
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
