0000174: puts for strings longer than INT_MAX

(0000271)
nick (manager)
2009-10-29 18:32

Discussed with some members of the C meeting after we adjourned ... general feeling is that this is undefined, and the EOVERFLOW error is an appropriate POSIX CX extension.

(0000272)
cmsmith (reporter)
2009-10-29 21:09

The return value for fputs() and puts() is "a non-negative number" upon success. The description does not indicate that the value needs to be related to the number of characters written.

(0000273)
drepper (reporter)
2009-10-29 22:58

Agreed, the value need not be the number of written bytes. But it'll break existing code to change that. Of course we can just return INT_MAX for all strings > INT_MAX. I guess this makes this a non-issue here.

(0000274)
geoffclare (manager)
2009-10-30 09:51

I agree no change to the normative text is needed, but I think we
should add some rationale or app usage about this issue. Perhaps
something like:

    "Traditionally puts() returned the number of bytes written.
    However, this convention cannot be adhered to for strings
    longer than INT_MAX bytes as the value would not be
    representable in the return type of the function. For
    backwards compatibility, implementations can return the
    number of bytes for strings of up to INT_MAX bytes, and
    return INT_MAX for all longer strings."

(0000298)
msbrown (manager)
2009-11-12 16:38

Add the following to the Application Usage section of puts(), p1723 after line 55077:

"In some implementations puts() returned the number of bytes written.
    However, this convention cannot be adhered to for strings
    longer than INT_MAX bytes as the value would not be
    representable in the return type of the function. For
    backwards compatibility, implementations can return the
    number of bytes for strings of up to INT_MAX bytes, and
    return INT_MAX for all longer strings."

Also the same on page 908 after line 30373, with puts() changed to fputs().

Issue History
Date Modified	Username	Field	Change
2009-10-29 18:21	drepper	New Issue
2009-10-29 18:21	drepper	Status	New => Under Review
2009-10-29 18:21	drepper	Assigned To	=> ajosey
2009-10-29 18:21	drepper	Name	=> Ulrich Drepper
2009-10-29 18:21	drepper	Organization	=> Red Hat, Inc.
2009-10-29 18:21	drepper	User Reference	=> ud-puts-overflow
2009-10-29 18:21	drepper	Section	=> puts()
2009-10-29 18:21	drepper	Page Number	=> 1723
2009-10-29 18:21	drepper	Line Number	=> 55058
2009-10-29 18:30	nick	Tag Attached: c99
2009-10-29 18:32	nick	Note Added: 0000271
2009-10-29 21:09	cmsmith	Note Added: 0000272
2009-10-29 22:58	drepper	Note Added: 0000273
2009-10-30 09:51	geoffclare	Note Added: 0000274
2009-11-12 16:38	msbrown	Interp Status	=> ---
2009-11-12 16:38	msbrown	Note Added: 0000298
2009-11-12 16:38	msbrown	Status	Under Review => Resolved
2009-11-12 16:38	msbrown	Resolution	Open => Accepted As Marked
2009-11-12 16:39	msbrown	Final Accepted Text	=> Note: 0000298
2010-08-27 13:05	ajosey	Tag Attached: tc1-2008

Notes
(0000271) nick (manager) 2009-10-29 18:32	Discussed with some members of the C meeting after we adjourned ... general feeling is that this is undefined, and the EOVERFLOW error is an appropriate POSIX CX extension.

(0000272) cmsmith (reporter) 2009-10-29 21:09	The return value for fputs() and puts() is "a non-negative number" upon success. The description does not indicate that the value needs to be related to the number of characters written.

(0000273) drepper (reporter) 2009-10-29 22:58	Agreed, the value need not be the number of written bytes. But it'll break existing code to change that. Of course we can just return INT_MAX for all strings > INT_MAX. I guess this makes this a non-issue here.

(0000274) geoffclare (manager) 2009-10-30 09:51	I agree no change to the normative text is needed, but I think we should add some rationale or app usage about this issue. Perhaps something like: "Traditionally puts() returned the number of bytes written. However, this convention cannot be adhered to for strings longer than INT_MAX bytes as the value would not be representable in the return type of the function. For backwards compatibility, implementations can return the number of bytes for strings of up to INT_MAX bytes, and return INT_MAX for all longer strings."

(0000298) msbrown (manager) 2009-11-12 16:38	Add the following to the Application Usage section of puts(), p1723 after line 55077: "In some implementations puts() returned the number of bytes written. However, this convention cannot be adhered to for strings longer than INT_MAX bytes as the value would not be representable in the return type of the function. For backwards compatibility, implementations can return the number of bytes for strings of up to INT_MAX bytes, and return INT_MAX for all longer strings." Also the same on page 908 after line 30373, with puts() changed to fputs().

Relationships

Aardvark Mark III