> timestamps isn't the only versioning mechanism for java webstart. it's just > the most fragile (as you already see) one. Why don't you use > maven-webstart-plugin? it creates webstart descriptor with artifact > version-based versioning.
We allready use maven to feed the jnlp, the problem is elsewhere. HTTP protocol used by java webstart use the If-Modified header. And if the say, log4j-1.2.14.jar is requested and the JWS cached version is earlier than the one in the webapp, it will be downloaded another time even it's exactly the same, HTTP protocol use time markers and not md5 or sha1 to difference between 2 files. > 2009/11/25, Henri Gomez <henri.go...@gmail.com>: >> >> >> When maven get the artifact from nexus to my local repository >> >> (~m2/repository), the timestamp is not conserved, the files get the >> >> timestamp of creation on the local repo. >> > >> > You are certainly welcome to file it as an enhancement, but it seems >> > like an edge case and I doubt it will be addressed unless you are >> > willing to do the work yourself and contribute it back... >> > >> > This is the only time I've ever heard of needing a particular >> > timestamp for a given artifact. I don't personally believe in there >> > being "one true timestamp" for a given artifact and think this should >> > rather be handled via proper versioning. >> >> >> Well it's more than needed when in some case like Java WebStart >> applications. >> >> Developper Joe get log4j from the repository on monday, log4j >> timestamp is monday locally. >> It produce an WS application on Tuesday where log4j is timestamped as >> monday. >> >> User Dan get the WS applications on Wednesday and the WS cache system >> mark the log4j (monday). >> >> Developper Jill get log4j from the repository on Thursday, log4j is >> timestamped Thursday locally. >> It produce an WS application on Thursday where log4j is timestamped as >> tuesday. >> >> Dan get back to the WS application on Friday, it's the same log4j but >> since the WS application report it to be Thursday and its cached >> version is earlier, it will download it another time. >> >> It just brake the WS/JNLP caching system and add uneeded network load. >> >> And if Dan is not alone, say 50 or 100 users, it became a serious >> infrastructure problem. >> >> I don't know very well the wagon support but I think maven could know >> the timestamp of the downloaded artifact ? >> If so, it should be easy to setLastModified after download ? >> >> Question, If I should file it as an enhancement, it should be for : >> >> 2.1.x, 2.2.x and 3.0.x ? >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: users-unsubscr...@maven.apache.org >> For additional commands, e-mail: users-h...@maven.apache.org >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@maven.apache.org For additional commands, e-mail: users-h...@maven.apache.org
