On 26/06/2013 16:13, Greg Stein wrote: > Hmm? Is it normal to put compiled outputs onto dist.apache.org > <http://dist.apache.org>? That .war file has a hojillion .jar files in > it. My understanding was that dist.a.o is for *source* distributions.
There are convenience binaries on dist.a.o but they are typically in the 10s of MB range. 2G appears to be completely OTT. > Further, it includes a bunch of SNAPSHOT artifacts in the multi-hundred > MB range. That seems like a total mistake. > > In addition, it contains TEN copies of poi-ooxml-schemas-3.8.jar (at 4M > each!). That is ludicrous. And TEN copies of netcdf, and TEN copies of > pdfbox. etc. You're just blowing away space with those multiple copies. > (I only looked at the three largest .jar files piled in there; I'm sure > you have ten copies of N other jars!) > > Why would you even WANT to put this onto our distribution system. Who > believes that a 2G distribution is remotely useful, natural, or proper? > The .war file seems full of improper bits, and then it gets hauled into > your other stuff, polluting them up to 2G. +1 > On Tue, Jun 25, 2013 at 10:21 PM, Lewis John Mcgibbney > <[email protected] <mailto:[email protected]>> wrote: > > Hi INFRA, > I did not realize that some of the Any23 release artifacts were in > the GB. > The following artifacts (around 7 of them) from the staging URL [0] > are in the GB range and I cannot commit them to SVN [1]. I also > remember seeing some traffic (a while back) regarding notice which > we were supposed to give you if we wished to publish artifacts of > this size... I therefore apologize for not getting this through to > you earlier. > Can you please suggest how we can release these artifacts? Right now, you can't. You have two options: 1. Justify why you need to have so many extremely large artifacts. If there is a valid justification, infra will figure out how to get them published. 2. Significantly (by at least one order of magnitude and preferably two) reduce the size of your release artifacts and roll a new release. Mark
