Hello,
I am getting an error when I attempt to install the "facets" gem. When I
enter
gem install facets
I get the following error message:
ERROR: While executing gem ... (Errno::EINVAL)
Invalid argument - llib/facet/nil/empty?
Are others seeing this (or am I guilty of user error)?
Brian
Hi Brian,
I just tested myself w/o error.
Since RubyGems is throwing the error and not Facets I'm not sure what
the cause is. Have you tried 'gem update --system' to update to the
lastest RubyGems version?
T.
Brian, am I right to guess that you are running Windows? Looks like
Facets has files called, for example, "empty?.rb", which is not valid
on Windows. I don't know of a way that you can get Facets to run on
Windows if that's the case, but perhaps its author has prior
experience making it work.
On 4/18/05, Brian Buckley <briankbuckley@gmail.com> wrote:
Hello,
I am getting an error when I attempt to install the "facets" gem. When I
entergem install facets
I get the following error message:
ERROR: While executing gem ... (Errno::EINVAL)
Invalid argument - llib/facet/nil/empty?Are others seeing this (or am I guilty of user error)?
--
Chad Fowler
http://chadfowler.com
http://rubygems.rubyforge.org (over 100,000 gems served!)
BTW, thanks for the Bug report.
Trans wrote:
Since RubyGems is throwing the error and not Facets I'm not sure what
the cause is. Have you tried 'gem update --system' to update to the
lastest RubyGems version?
I have the most current RubyGems and I have the same message installing Facets.
Cheers,
--binkley
Thanks T.,
At your suggestion I ran 'gem update --system' but unfortunately that didn't
resolve my problem -- I am still seeing the same error.
FWIW I am using RubyGems for a handful of others gems without trouble. In
case it matters I am on Windows XP SP2, Ruby 1.8.2.
Brian
Ah. Thanks Chad. That was what I was starting to conclude myself.
I've tried to keep the file name as close as reasonably possible to the
main method name(s) of the facet. But I know that RDoc via Rake also
chokes on a couple of the names. I think that all modern file systems
now allow fully arbitrary file names (Although there may yet be some
limitations on this I am unaware of.). So the appearent reason is that
quotes aren't being used by programs when they shell out commands. For
instance, the invalid argument above should probably be:
-l"lib/facet/nil/empty?"
Chad, would that fix the problem?
Thanks,
T.
What platform are you running?
T.
Trans wrote:
I think that all modern file systems
now allow fully arbitrary file names (Although there may yet be some
limitations on this I am unaware of.).
If that is the case, Windows XP still lacks a modern OS. 
Several characters are forbidden, among them the question mark, asterisk and colon. Some of the restrictions are in NTFS and the kernel, some are part of the Win32 API subsystem.
Cheers,
--binkley
You would be wrong.
Windows filesystems do NOT allow a long series of characters
(/:?\<>).
Anyone who names a *file* with a question mark or an asterisk should
simply be shot.
As should anyone who has both a README and a Readme or a Makefile
and a makefile.
-austin
On 4/18/05, Trans <transfire@gmail.com> wrote:
Ah. Thanks Chad. That was what I was starting to conclude myself.
I've tried to keep the file name as close as reasonably possible
to the main method name(s) of the facet. But I know that RDoc via
Rake also chokes on a couple of the names. I think that all modern
file systems now allow fully arbitrary file names (Although there
may yet be some limitations on this I am unaware of.). So the
appearent reason is that quotes aren't being used by programs when
they shell out commands. For instance, the invalid argument above
should probably be:
--
Austin Ziegler * halostatue@gmail.com
* Alternate: austin@halostatue.ca
Trans wrote:
What platform are you running?
Yah, the empty?.rb not a valid file name under Cygwin (Windows) gives the show away.
Cheers,
--binkley
Anyone who names a *file* with a question mark or an asterisk should
simply be shot.
Sorry Austin, but that's just a completely asinine statement. First of
all, some might say such things about method names too! --which is what
these file names are based on. Secondly, let's back up a couple decades
and try statements like, "Anyone who names a *file* with more then 8
characters should simply be shot." You may think of filenames as
nothing more the special grepable_computer_ id keys, but the reality is
_people_ have to use them to give documents recognizable names. But
alas as far as I'm concerned we're still in the dark ages when it comes
to managing our data. Why it's not stored in RDBMS yet is beyond me.
Anyhow, this is taking us away from the real issue.
Thanks for the info on the limitations of Window's filenames. I did not
realize they still existed.
So now I have to change the file names. Great! Well, I'll let you all
know when the fixes are uploaded.
Any suggestions on naming these is welcome.
T.
Austin Ziegler <halostatue@gmail.com> writes:
Windows filesystems do NOT allow a long series of characters
(/:?\<>).Anyone who names a *file* with a question mark or an asterisk should
simply be shot.
As if spaces would be any better, but I really don't want to get
deeper into that.
On 4/18/05, Trans <transfire@gmail.com> wrote:
As should anyone who has both a README and a Readme or a Makefile
and a makefile.-austin
--
Christian Neukirchen <chneukirchen@gmail.com> http://chneukirchen.org
Trans wrote:
Any suggestions on naming these is welcome.
At least for the cases of foo?-type methods I'd suggest the Schemish convention of the -p suffix ('p' here stands for predicate), e.g., foo? => foo-p. It should work on all platforms and reflects the heritage of Ruby method names such as using the bang (exclamation mark) suffix for mutators.
Cheers,
--binkley
Thanks Binkley. I'm working on a solution now. So you're suggesting I
use '-p' for '?' and '-x' for '!', I take it? I'm debating either that
or just dropping the '?' and '!' from the file name altogether. I
figure if there are multiple methods of the sort #x and #x? they ought
to be related enough to coexit in the same "atomic" facet file.
The other one I'm not sure about is '[].rb'. Windows does accept this
as a valid file name, but I worry perhaps some other systems may choke
on it. I'm just not sure what to rename it.
Thanks,
T.
Trans wrote:
Thanks Binkley. I'm working on a solution now. So you're suggesting I
use '-p' for '?' and '-x' for '!', I take it? I'm debating either that
AFAICT the exclamation point is actually ok for filesystems. It's the question mark which draws trouble, curled in a cipher as it is. So my suggestion, pending validation on all Ruby platforms, is:
foo! => foo!.rb
foo? => foo-p.rb
Cheers,
--binkley
B. K. Oxley (binkley) wrote:
foo! => foo!.rb
foo? => foo-p.rb
Of course, a more general and possibly practical suggestion rather than a table of special cases is to translate punctuation marks into their ASCII value:
foo! => foo-21.rb
foo? => foo-3f.rb
=> -5b-5d.rb
Or some similar scheme with a suitable escape marking character (yes, using the minus sign is kind of lame). But I might be drifting in the land of excess engineering.
Cheers,
--binkley
B. K. Oxley (binkley) wrote:
foo! => foo-21.rb
foo? => foo-3f.rb
=> -5b-5d.rbOr some similar scheme with a suitable escape marking character (yes, using the minus sign is kind of lame). But I might be drifting in the land of excess engineering.
That only works if you don't see any problems telling people that if they want the redefined methods then they need to type
require "-5b-5d"
--
Mark Sparshatt
mark sparshatt wrote:
That only works if you don't see any problems telling people that if they want the redefined methods then they need to type
require "-5b-5d"
Can you not teach require to translate the file names? (I'm too new to Ruby to know the answer to that. I know I can do it in Perl.)
Cheers,
--binkley
B. K. Oxley (binkley) wrote:
mark sparshatt wrote:
That only works if you don't see any problems telling people that if they want the redefined methods then they need to type
require "-5b-5d"
Can you not teach require to translate the file names? (I'm too new to Ruby to know the answer to that. I know I can do it in Perl.)
This seems to work, though there's probably a better way to do it
class Object
EscapedChars = ["[", "]", "?", "!"]
alias :old_require :require
def require(name)
EscapedChars.each do |char|
name = name.gsub(char, "-#{char[0].to_s(16)}")
old_require name
end
end
require "" #=> translated to require "-5b-5d"
--
Mark Sparshatt