no, I just expressed my opinion, which was that: 1. read access should be committed to CVS before write access. 2. write access makes sense when it's more or less complete (e.g. if you add you should also be able to delete it properly; otherwise you add a record, then delete a record and the two files - database and memo file - suddenly fall out of sync. Not good. Better not to touch things than corrupt things. Or did I miss something?) 3. we should be able to specify the extension for the memo file (almost feels like you need an optional parameter for dbase_open() and dbase_create() - shudder...)
But then again, it's just my feedback based on my experience with doing the same stuff you are doing now. You can post your changes here (unified diff) and if someone feels that the code is sane and won't break things, they are more than welcome to commit it. Vlad Geoff Jukema wrote: >You'd be suprised how many people are actually using dbase as a common-file >for transferring small chunks of data - this is why I needed to write the >write code for our company. > >The dbase_open() and dbase_close() work in the 'buggy' version that reads >memo data. I can add it to the current 'released' version if you think it >should be there. > >You are absolutely correct about the fpt vs dbt. We required foxpro files, >so I wrote it to handle those. In the README.MEMO file, it describes what >is needed for handling dbt's, other than actually replaceing the "fpt" >occurrences with "dbt" - I should come up with a nicer way of handling >this - perhaps a constant with an option. Any thoughts here? > >So, what I think your saying is that even though I have the >dbase_add_record() completed, before it's committed, I should get the read >support completed? At this point, it could be committed. > >Thanks, >Geoff >----- Original Message ----- >From: "Vlad Krupin" <[EMAIL PROTECTED]> >To: "Geoff" <[EMAIL PROTECTED]> >Cc: <[EMAIL PROTECTED]> >Sent: Tuesday, May 21, 2002 10:28 AM >Subject: Re: [PHP-DEV] dbase extension - Contributing Code > > >>this would be nice to have - I've written some code for that a while >>back, but never got to finishing or committing it, so it's all abandoned >>now. >> >>I briefly looked at your code and have a couple of comments. First, >>writing to dbase is optional for most people (because it's not safe to >>write from multiple instances of PHP running, and you are just asking >>for trouble that way anyway) but reading from it is, indeed, useful to >>retrieve some legacy data. >> >>for starters, it would be nice to have dbase_open(), dbase_close() (they >>should probably check dbf_dhead.dbh_dbt - see dbf.h - this field already >>gets filled when you do dbase_open() on a file). >> >>...which brings me to another question - you are using .fpt extension >>for memo files, right? wasn't it .dbt or something like that? I have a >>nagging suspicion that .fpt is what FoxPro decided to use, and everyone >>else uses something else. >> >>after that some read-only support would be nice (dbase_get_record() / >>dbase_get_record_with_names() pair). >> >>after that's ready, we can add some write support, but when it's added >>it would be nice to add changes to dbase_add_record() together with >>changes to dbase_delete_record() and dbase_pack(). It's sad if we can >>add but not delete. I'd rather have read-only access than a handicapped >>write access. >> >>If you submit a unified diff against the HEAD branch to this list, >>chances are someone will add it (as long as it does not obviously break >>things for other people). >> >>Those are just a few thoughts... >> >>Vlad >>-- >>Vlad Krupin >>Software Engineer >>echospace.com >> > > > -- -- Vlad Krupin Software Engineer echospace.com -- PHP Development Mailing List <http://www.php.net/> To unsubscribe, visit: http://www.php.net/unsub.php
