Re: package badness

Andy Piper <andyp(a)parallax.co.uk&gt; writes in xemacs-beta(a)xemacs.org: Even if true, it is no excuse for XEmacs package installation to puke its guts out. Packages differ wildly in size. The top 4 account for about 33% total disk usage. Depending upon what has been installed, that may be higher. For me, I see 71MB in a fully bytecompiled xemacs-package tree and 53MB (39.2MB XEmacs, 13.8MB Mule) for what I have installed[1]. The "Big 4" packages: Gnus 6.3MB Leim 6.0MB Calc 5.6MB SKK 5.6MB = 23.5MB Writing as an expert in filling up disks of any size (:-), I repeat my contention that we have to take into account filling disks and doing something graceful when that occurs. I have blown out the disk I use for distribution building on numerous occasions. To complete summarization of disk usage. For building I see the following: 71MB XEmacs packages as of last night 45MB XEmacs 21.0-pre9 43MB XEmacs 21.2.4 For installation: 21.5MB Core 23.5MB "Steve" XEmacs package installation[2] 63.6MB Full XEmacs package installation For comparison, XEmacs 20.4: 50.1MB (Compressed Lisp) 67.6MB (Straight Installation) Based on the kind of software I have installed, I don't know whether I'm relieved or disappointed that XEmacs is no longer anywhere close in the running for largest software package I have installed. The record is held by Oracle database server which is asking for 700MB just for installation not counting the "at least 3" mount points it wishes for storing database stuffs. Footnotes: [1] I have all of the "Big 4" installed, so that's 44% for me. [2] I like very much the fact that without any Lisp .el compression, I have a 10% smaller installation of 21.0 with more features that _I want_ than 20.4 with Lisp .el compression.