Hello Gabriele > Hello [EMAIL PROTECTED]! > > On 06-Mag-00, you wrote: > > a> the "REBOL-header" is skipt even if it is somewhere in the > a> code, even in nested blocks?! > a> when at start of line! > a> this means i have to check for occasional REBOL everywhere in > a> my save - databases?! > a> Bug, yes? > > No, I think this is intended behaviour. (rip works thanks to this, > too.) If the string contains a REBOL header, it is assumed to be a > script... but perhaps this should happen for files only, and not > for strings? > Well, rebol talks about simple databases and "better xml". how can i use it if a occasianal "rebol" can crash the data? . REBOL can be part of the data on save and after that load will crash, in an otherwise correct programm. maybe a new refinement LOAD/DATA which ignores headers? Or an explicit option at start for skipping before REBOL, perl does it this way. i think this header-trap is as random as dangling pointers, if one does not know about it. This stuff should be avoided in a beginners language. Or there should be a warning: "currently not for saving serious data!" for me i was expecting i can load/next parts of rebol-source. i loaded the header-block by hand to get only defined fields, after moving behind the first rebol i used load/next . works very well, except with the parser itself. load moaned, and i could not figure out why. only occasinally i discovered the header-way of load/next, and the parser contains a REBOL somewhere.. BTW i also dislike the [a: ""] trapp. i programm the usual way, "write, test run, change, test .." without restarting rebol. and if the test works fine, i'm shot in the back if a second complete run gives "random" errors. sourcecode should try to show what will happen, without this "oh yes, learn the magic rule #44b23, says [a: copy""]! stupid boy!" especially in rebol, which seems so clear i ride with lowered shield.. I suggest inline-strings and blocks should be read-only, copy [] can be written. this should fix surprises. and COPY/PROTECTED could be a feature? I heard this [a: ""] was discussed some time ago, is there a collection of arguments somewhere? Then there must be a timeout if rebol runs serious remote (cgi..). And a warning about contexts/bind while this crashes so reliable. What else? Is there a public-readable bug-list somewhere? Because, avoiding some stuff, rebol is very stable, and the real trapps could be mentioned on a single page? There should be one. I would not trust guys shooting me in the back once a long time, (except i like them so mutch :) A longer time than not to try because i read helpfull warnings about it, maybe.. > Regards, > Gabriele. > -- > Gabriele Santilli <[EMAIL PROTECTED]> - Amigan - REBOL programmer > Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/ > > Volker >