Re: bug in ed-0.3/ed-0.4

[Andrew L. Moore]

Antonio,
What happened to the new version of ed that I gave you last summer
which you were so eager to clean up for distribution?  After a
cursory look at "your" code, what I see is ed-0.2.  In the handful
of places where you took the liberty to change ed 0.2, you introduced
dangerous errors.  A few examples (far from complete):

 1) Your code breaks on NFS file systems.  Hint: don't touch open files.
 2) It may look pretty to your aesthetic, but a comparison such as
              if (big_uint_a > 2* big_uint_b) ...
    will overflow.
 3) C-d on stdin is not handled properly, or at all in some cases.
 4) File I/O is broken.

Perhaps it is time to cut your loses and withdraw your work from
circulation. Say the word, and I am happy to take over your difficult job.
-AM


] I do not think ed is a joke. I think ed is a program I do not use, a 
] program I do not like, and a program I am voluntarily maintaining to 
] help the GNU project just because nobody else wants to maintain it.
] 
] The '-' argument of ed is a bizarre thing, but AFAIK '%' is simply 
] redundant. I won't add code to the already complex definition of address 
] ranges in ed to support this. Much less make it depend on 
] POSIXLY_CORRECT. I do not have the time to do this. Slavish adherence to 
] past releases' behaviour is specifically not a goal for me.
] 
] I do not want to impose anything to anybody. If it is true that ed has a 
] community, it will be easy to find somebody willing to maintain it. Or 
] perhaps Redhat could pay someone (not me) to maintain ed. Maybe even one 
] of you is able to give a reason beyond backward compatibility to support 
] '%'. In the meantime I will only fix "real" bugs in ed.