Hi all!
Some weeks ago, I volunteered for beeing the archiver at the upcoming GPC archive. In the following, I shall lay out the prinziples and structures I would like to follow. All suggestions and comments are welcome.
**1** general policy The archive will contain * the gpc-source and supplied binaries * related files (i.e. the gcc-source) * contributed tools/apps/libs The main policy should be "less is better". This means that we would prefer a well organized archive with a content that is truly spezific to gpc over an archive which contains much misc packages which could also be found in other archives.
**2** Incoming policy This leads to the point, that packages which are to be included, should be selected from the archiver, rather than from the author or uploader. There will be an "Incoming"-directory with write-access. The archiver will select the appropriate files and add them to the archive. All uploaded files should contain a file foobar.lsm (which should be really descriptive not just "it is a library for graphics") and can contain a file foobar.readme. Templates will be supplied, of course. But thats all not that critical, as far as the archiver is human and not some script.
**3** Access Access to the archive should be possible via raw ftp as well as via the comfortable www (but viewable with any browser, including lynx). The best solution should be a ftp-directory with files "index.html" in each directory. Some may be familar with this from the sunsite.unc.edu linux-archive.
**4** The files index.html They contain * subdirectories with description * files in the current directory with description * links to navigate inside the archive : UP NEXT PREV ROOT GPC-HOME * maybe a text with information related to the current directory I would like to have the files index.html be very customized - i.e. descriptions could include links to the homepage of the package, the text could provide links to related interesting places. I personally dont like scrolling, if Im working in an archive. So I would try to design in a way, that at least the subdirectories are within one page, better that subdirectories+files are within one page. A typical page would look like ___________________________________________________ PREV NEXT UP ROOT GPC-HOME
This is the Source of the Version 2.5.76 of GPC. MORE /\ /\ || || Link to a tree, see 5 Link to the Text at bottom
subdirectories with description files with description
some text (explanations, links) _____________________________________________________
**5** Other navigating files As a second concept of navigation, there will be a tree (<ul>) representing the archive. The Name of the directory ("This is the Source of the Version 2.5.76 of GPC" above) would link to this tree. For raw ftp-access there will be an index.txt.
**6** tree structure of the archive --stable-----binary---linux .tar.gz and .rpm ask for contributed .deb | |-freebsd | |-...(djgpp, solaris) ask for contributions |-source .tar.gz, .src.rpm, gcc-source, ask for source-deb --unstable---versionx Source of unstable versions (beta versions) |-versiony If it doesnt make sense, put all in one |-versionz directory
--old--------versionx Dito |-versiony |-versionz --doc list, homepage, faq, fpkvsgnu --devel gpcasm, solaris-includes, tply wenn portiert, bpcompat, bgi2grx, xbgi --contrib relochtml, lord --unported what is worth porting
--system-----djgpp Directories made via symlinks. The user goes |-linux-redhat into the archive, directory |-linux-debian system/$HIS_SYSTEM. There (s)he will find |-linux-targz current stable/unstable source/binary |-emx |-freebsd |-...other. ask for contributions
--Incoming
**7** Search A searchfunction will be included in future. It should do search for category, name, word from name, description, lsm-file. Search-sheme would be regexp, substring, exact string, case sensitive/insensitive, agrep(fuzzy-search)
**8** Other future Plans Not much. Any Ideas? Of course if the archive gets to big, one must set up an automated archiver, based on the foobar.lsm-files.
These are Peters and my ideas about the archive. I hope I could make the general idea clear. Any suggestions and comments are really appreciate.
As I will have no internet-access in march, maybe I wont get it done till the first days of april.
Hans