[Prev][Next][Index][Thread]
Re: netatalk-style resource forks? (please? :)
>>>>> "Brad" == Brad Midgley <brad@pht.com> writes:
Brad> hello... desktop representation is out of the question (too
Brad> different for everybody), but executor should at least put
Brad> its resource forks in a subdirectory without mangling the
Brad> name. Other schemes could depend on softlinks to that
Brad> directory. (would only have to be fixed up when folders
Brad> were created) Mangling the name in the current directory is
Brad> ugly.
The problem is that the Apple Double spec. is an incredibly bad
specification, but rather than go our own way, we adhere to it. There
are actually a few other Mac connectivity tools that adhere to the
spec. as written and Executor will interact with them correctly (off
the top of my head I think InterCon has a product that does this and I
think Paul Hargrove's HFS Linux filesystem can create in-spec
AppleDouble names).
Brad> The easy road to get executor to deal with a network on a
Brad> netatalk-supported host will be to depend on kernel DDP and
Brad> link in the netatalk libs. so why not use a compatible
Brad> naming scheme now and avoid the headache later?
Yes and no. We could make a very quick hack to support the netatalk
style names and have considered doing it at least for our own internal
use (Cotton uses netatalk) but have so far resisted. I think Cotton
plans on writing a little script that will recursively convert a
directory and its children in either direction between the netatalk
naming convention to the AppleDouble naming convention. That will be
useful to him and perhaps to many of you out there as well.
Internally Executor and netatalk both use the AppleDouble file format;
it's only the naming where they deviate from the spec.
Brad> btw, ardi's bugs/feedback page is broken. anyone there?
Send e-mail to "webmaster@ardi.com" and cc it to "bugs@ardi.com" with
more detail and we'll get it fixed ASAP. We have received a few bug
reports here and there so we assumed it was working.
Brad> On Thu, 12 Oct 1995, David E. Hollingsworth wrote:
>> In article
>> <Pine.LNX.3.91.951012081238.1368B-100000@exodus.pht.com> Brad
>> Midgley <junkmail@pht.com> writes: > I'm not sure if this has
>> come up yet, but since executor doesn't have > network support,
>> could it at least support the netatalk method of keeping >
>> resource forks in a .AppleDouble directory? (how about a
>> non-defaulted > command-line option?)
>>
>> Of course, CAP aufs uses .resource directories, Helios
>> EtherShare and IPT uShare use .rsrc directories, Xinet KA-Share
>> uses .HSResource directories, and Pacer PacerShare uses
>> afp_resource directories.
>>
>> There are corresponding differences for desktop representation.
>> Many of these systems also appear to have files or directories
>> for storing finder information. Fun, eh?
Brad> brad@pht.com
References: