Issues:

  1. controlling editing access to prevent spam and vandals. yes.
  2. permanent web address. set up dynamic dns with a cool name. several mirrors with boring names.
  3. storage requirements: 1-2 TB for the first year for each location (main and mirrors). 10 TB/location in the first few years.
  4. how to store and index our data. metadata: README file-who, what, algorithm, computer algebra system, operating system, when, timing,level of rigour and precision, description of datafiles, tex okay. data should reside in ascii, machine friendly format (easy to parse. no extra junk such as html). should not be affected by adding to the directory. David's demo- do we allow other contributors to add to a person's directory? index L-functions according to selberg data.
  5. upload should be a directory which has: README, a directory for data, and directories (optional) for code and papers.
  6. .code.tar as an optional record of what was used for the computation. copyright issues.
  7. access/procedure for uploading data. decision: refereeing process for data.
  8. ability to locally search through data and extract subdatasets. relational databases? cgi grep script?
  9. standards to allow (require by agreement!) referencing in papers of data used. copyright discussion.
  10. too many files? not more than n files. SQL databases might help.
  11. modularity of code and data.
  12. md5 sums.

LfunctionsAndModularForms/Design/Decisions (last edited 2009-03-01 00:55:49 by localhost)