bobm@rtech.UUCP (Bob Mcqueer) (09/04/86)
[]--- Concerning the "#W" lines in map files: it seems to me that the "date" portion of that line is a very useful piece of information, but non-adherence to format is an even worse problem here than with the "#L" lines recently discussed. Looking over the map files, I see: 1) "date" format beginning with day-of-week 2) "date" format beginning with month (no day-of-week) 3) "date" format beginning with numeric day of month 4) six digit integer YYMMDD preceding semicolon 5) 4) following semicolon or with no semicolon. 6) MM/DD/YY at some random placement 7) variations with and without time-of-day portions The above appear in a significant number of entries. There are also many entries which convey the desired information to a human reader, but don't adhere to any of the above formats. Please note: it is not my intention AT ALL to chastise anybody for not entering the dates "properly". I'm just noting that as things stand now, it is difficult to write a routine which will reliably pick the date out of most of these lines. I was wondering. Could part of the collection process for maintaining mapfiles and building the distributed archives be attachment of a time stamp to each entry indicating when it was received? The date stamp would then be in a consistent format. True, this hides the fact that somebody may have mailed the map maintainer an ancient entry, but I think the guarantee of being able to machine parse the date is an equitable trade-off. Another thought: there ought to be a little program floating around which would check the validity of map entries, including the format of lines which could reasonably be machine readable. Is there such? If so, making the tool available and known would help matters. If everybody had the tool, and it produced explicit diagnostics, it might become reasonable for the map maintainer to insist on proper format before including entries in the archives. Bob McQueer -- {amdahl, sun, mtxinu, hoptoad, cpsc6a}!rtech!bobm
tgt@cbosgd.UUCP (Tim Thompson) (09/04/86)
In article <431@rtech.UUCP>, bobm@rtech.UUCP (Bob Mcqueer) writes: > Concerning the "#W" lines in map files: it seems to me that the "date" > portion of that line is a very useful piece of information, but non-adherence > to format is an even worse problem here than with the "#L" lines recently > discussed. Looking over the map files, I see: > > [ formats omitted for brevity ] > > The above appear in a significant number of entries. There are also many > entries which convey the desired information to a human reader, but > don't adhere to any of the above formats. > Bob McQueer > -- > {amdahl, sun, mtxinu, hoptoad, cpsc6a}!rtech!bobm Bravo, Bob! I'm the map maintainer for Indiana, Illinois, Pennsylvania, Ohio, North Carolina, and Tennessee. I just took over the job, and before I did anything else, I went through by hand and standardized the "#W" line for every single entry for each state. It was a TIME-CONSUMING job, but now that it's done, all I have to do is make sure that incoming updates adhere to the standard. Let me quote from the document that (from what I understand) comes with the Usenet software: The entire map is intended to be processed by pathalias, a program that generates UUCP routes from this data. All lines beginning in `#' are comment lines to pathalias, however the UUCP Project has defined a set of these comment lines to have specific format so that a complete database could be built. [...] #W who last edited the entry and when This field should contain an email address, a name in parentheses, followed by a semi-colon, and the output of the date program. Example: #W ucbvax!fair (Erik E. Fair); Sat Jun 22 03:35:16 PDT 1985 The same rules for email address that apply in the contact's email address apply here also. (i.e. only one system name, and user name). It is intended that this field be used for automatic aging of the map entries so that we can do more automated checking and updating of the entire map. See getdate(3) from the netnews source for other acceptable date formats. [ END OF QUOTATION ] From this piece of documentation, I got the standard that I'm going to use for these six states. So right now, if I were to update the entry for the site cbosgd, the entry would look like this: #W cbosgd!tgt (Tim Thompson); Thu Sep 4 13:42:14 EDT 1986 I never got a chance to see getdate(3) for other acceptable formats, and besides, it seemed easier to use the output from an an existing utility than to have to format the date by hand. > Another thought: there ought to be a little program floating around which > would check the validity of map entries, including the format of lines > which could reasonably be machine readable. Is there such? If so, making > the tool available and known would help matters. If everybody had the tool, > and it produced explicit diagnostics, it might become reasonable for the map > maintainer to insist on proper format before including entries. I'm currently gathering some thoughts on a program that would provide the functions you mention. If such a beasty already exists out there on the met somewhere, someone PLEASE send me e-mail, before I go ahead and re-invent the wheel. > Bob McQueer > -- > {amdahl, sun, mtxinu, hoptoad, cpsc6a}!rtech!bobm Tim Thompson cbosgd!tgt -- ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Timothy G. Thompson AT&T Network Systems Columbus, Ohio cbosgd!tgt DISCLAIMER: These ramblings are my own. However, a thousand monkeys pounding on a thousand typewriters would eventually produce the exact same thing!! +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++