Re: BXE Wishlist
Emmet Caulfield <[email protected]> Sun, 21 Nov 2004 17:22:13 -0500
| Newsgroups | gmane.editors.bitflux.general |
|---|---|
| Organization | Netrogen.com |
| Message-ID | <[email protected]> |
On Sun, Nov 21, 2004 at 09:33:46PM +0100, Christian Stocker wrote: > Sure, but BXE shouldn't send empty files or documents just consisting of > the root node. You can't produce empty XML files with BXE at all > (currently), but the later would be possible ;) > > To sum up, BXE shouldn't send stuff over the wire, if an error happened. > I don't know right now, why this happens at all. But Jean Pierre will > certainly shed some light on that. And yes, you definitively should also > do some checks on the server side. The point that I was trying to make was that, in spite of the fact that BXE should, as you say, make a best effort to supply only valid data to the backend, BXE mustn't be held responsible for overwriting valid data with invalid data. The responsibility for ensuring that valid data is over-written only with more valid data can only properly rest with the over-writing application, not the supplier of the data, be it BXE or otherwise. Selfishly, I just want to make sure that you're spending your time improving BXE and adding great new features, not pursuing the impossible ;o) When BXE (or anything else) "misbehaves", we would do well to assume that it will do the worst thing that is possible for it to do (like the famous buttered toast always landing butter side down :o), and defend ourselves against that possibility; perhaps for BXE this "worst thing" is sending back an empty document? I'm not altogether sure that it's possible to write a program that has such perfect self-knowledge and well-defined behaviour under error as "never sending back empty files or document just consisting of the root node" would require. Even if it is, there will always be a slim chance that the back-end will get invalid stuff through a Byzantine error on the part of BXE, the browser, the web-server, either host OS, or (in extremis) some network device between client and server. Anyway, enough of me waffling: I think we agree that the only sensible thing to do is to check on the server-side. Cheers, Emmet. -- ,----------------------=[ [email protected] ]=------------------------. | Internet Software & Business Automation. C, Perl, XML, SQL | | Legal: http://netrogen.com/about/email.php Unix/Linux, Windows | `---------------------=[ http://netrogen.com/ ]=-----------------------' -- bx-editor-users mailing list [email protected] http://lists.bitflux.ch/cgi-bin/listinfo/bx-editor-users