This is the mail archive of the crossgcc@sourceware.org mailing list for the crossgcc project.

See the CrossGCC FAQ for lots more information.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: Cygwin and crosstool! (vmlinux.lds.asm problem with 2.6.12.5 case-insensitive build patch from crosstool)


----Original Message----
>From: Dan Kegel
>Sent: 27 August 2005 01:45

> Dave Korn wrote:
>>> Corinna objects to it, though; she said on 22 Jul 2004 22:46:44 in
>>> a message in thread "Re: Slight problem with case sensitivity on managed
>>> mounts with C VS-1.1" archived at
>>> http://www.newsarch.com/archive/mailinglist/cygwin/msg09026.html :
>> 
>> 
>>   I can't get that address to resolve!
> 
> Search for the content in google; it's still in the cache.

  Ah, it's also at the canonical location of
http://sources.redhat.com/ml/cygwin/2004-07/msg00909.html

<koff>  Did you notice who that post was a reply to?


>>> But maybe she'd be ok with it as an extra mount flag.
>> 
>>   I hope so; that way it's strictly isolated from anyone who doesn't
>> want it in use.  Making it a property of a mountpoint (rather than e.g.
>> a CYGWIN environment variable option) should make it keep itself to
>> self-contained dir subtrees, rather than allowing chaotically-named
>> files to scatter across the whole fs.
> 
> Not the whole fs; just the part mounted as 'managed'.  And
> the mangled names are only visible if you don't look at
> them through cygwin, of course.

  Nonono; I was referring to the only-discriminable-by-case files that my
new option would create; if it was a CYGWIN environment variable flag that
enabled _POSIX_ flag globally, these troublesome files could be created
arbitrarily all over the fs, where it would only be a matter of time until
they ran up against case-insensitive win32 apps with bad consequences; hence
my reasoning to make it a mount mode and, just like with mangaged mode, have
all the files with wierd names in one subtree.

>>> But before you do: what *is* the overhead of managed mode mounts?
>> 
>>   No idea, but I'm sufficiently off-put by those mangled names that I
>> don't like it!
> 
> Aha.  Then please don't argue that there is any overhead.
> Your real objection, the mangled names, is weakened if you
> tie it to unproven and possibly false statements.

  Oh come on!  The code paths for managed mode are exactly the same as the
code paths for non-managed mode, with the addition of extra function calls
such as fnunmunge and special_introducer and so on, so I believe my
statement is in fact trivially demonstrable by inspection.  Read the source!
My objection is that there is _an_ overhead *and* it looks ugly.  I'm not
proposing to remove or replace managed mode, so I don't think I need to
comprehensively demonstrate the size of the overhead. :)


    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....


------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]