Conversion between C string and std::string lead to garbled text in platform specific way

Question

I was dealing with a platform specific bug today in which on a Windows machine a certain string would be quite garbled, but not on a Mac. The bug had to do with several lines that did both explicit and implicit conversions between std::string and const char *. Basically, I had a function with the signature

void foo(const std::string &id);

where foo at some point prints the string. On Windows, if called like below it would print the id string with various levels of corruption (garbling the first few characters or as much as the whole string)

std::string mystring = bar();
const char *id = mystring.c_str();
foo(id); // pass the C style string in because I thought that's what it took

I corrected the error by calling foo correctly:

std::string mystring = bar();
foo(mystring);

I can't figure out a few things though, like

What was the source of the bug?
Why was it platform specific?
Is implicit conversion between const char * and std::string ever safe?

The third point's answer is not when it's a null string or not pointing to a valid string. — Qaz
– Qaz, Commented Jul 18, 2013 at 21:37

Mark Ransom · Accepted Answer · 2013-07-18 22:25:30Z

4

I'm going to take a wild guess and say that the code in your question isn't representative of your real code. The symptom sounds like memory being reused, which means the string isn't valid anymore at the time of the foo call. You could get that by using c_str on a temporary, for example const char *id = bar().c_str();.

answered Jul 18, 2013 at 22:25

Mark Ransom

310k44 gold badges423 silver badges660 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Sevenless Over a year ago

Totally right. In writing the question I changed the code removing bar().c_str() because I didn't think through the fact that the return from bar would be temporary. Thanks!!!

David Schwartz · Accepted Answer · 2013-07-18 21:48:44Z

2

There are two likely possibilities, and from just what you've said, we can't tell which:

foo mishandles the case where it gets a reference to a temporary. For example, perhaps it stashes a pointer to its parameter somewhere and other code uses it after foo returns.
The data in mystring contains embedded zero bytes and the length is lost when you call c_str.

answered Jul 18, 2013 at 21:48

David Schwartz

184k18 gold badges229 silver badges293 bronze badges

3 Comments

Mooing Duck Over a year ago

Hadn't thought of #2.

Mark Ransom Over a year ago

Embedded zero bytes would cause truncation, not garbling.

David Schwartz Over a year ago

@MarkRansom: Truncation can cause garbling. For example, consider if the last byte in the string indicates its encoding format. Truncation would mean the last byte would be changed, causing the encoding format to be misunderstood, causing garbling. Or consider if multibyte is the default unless the string is an odd number of characters ... again, truncation would cause garbling.

user539810 · Accepted Answer · 2013-07-18 22:14:38Z

0

I don't know what is going on, but is there any reason not to copy the result of c_str() into a buffer you own, call foo() on it, and destroy the buffer you created? If that doesn't fix it, it is a problem with foo() only rather than the implementation or some conflict between the implementation and foo().

answered Jul 18, 2013 at 22:14

user539810

Collectives™ on Stack Overflow

Conversion between C string and std::string lead to garbled text in platform specific way

3 Answers 3

1 Comment

3 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related