Re: Fonts reorganization and additions

[Ken Brown <kbrown@cornell.edu>]

On 9/13/2015 12:10 PM, Ken Brown wrote:
> On 9/11/2015 3:59 PM, Yaakov Selkowitz wrote:
>> The first thing that should be checked is if the code in setup which
>> allows it to unpack files with "illegal" characters in Windows also
>> makes sure to create parent directories.  Based on these reports my
>> *guess* is it doesn't but I haven't looked yet.
>
> I'm not very familiar with the setup code, but I think I may have found
> the problem.  The function mkdir_p() in mkdir.cc contains the following,
> starting at line 99:
>
>    for (c = path; *c; c++)
>      {
>        if (*c == ':')
>      slash = 0;
>        if (*c == '/' || *c == '\\')
>      slash = c;
>      }
>
>    if (!slash)
>      return 0;
>
> If I understand this correctly, the function will return success without
> having created the leading directory if the path name contains a colon
> and doesn't contain a slash after the last colon.  I guess the
> assumption is that there is no "leading directory" to be created in this
> case.  [This part of the code was written in 2000, when a colon could
> not be part of a file name.]
>
> I hope someone who understands the setup code better than I do can take
> a look at this.  I'm not confident that I understand all the uses of
> mkdir_p() or all possible reasons why its in_path argument might contain
> a colon.

The following patch fixes the problem:

diff --git a/mkdir.cc b/mkdir.cc
index dabc256..84bb176 100644
--- a/mkdir.cc
+++ b/mkdir.cc
@@ -98,8 +98,6 @@ mkdir_p (int isadir, const char *in_path, mode_t mode)

    for (c = path; *c; c++)
      {
-      if (*c == ':')
-       slash = 0;
        if (*c == '/' || *c == '\\')
         slash = c;
      }

I don't know if this might cause some other problem, but it seems OK in 
limited testing.

Ken


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple