Bo Peng wrote: > I have to say that, if this is indeed boost::iostreams' fault, I would > be really disappointed at the quality of boost libraries.
There is no single 'quality of boost libraries'. There are excellent libraries, but there are also useless or complicated ones. And there are unmaintained libraries (like iostreams). BTW, did anybody file this as a boost bug? > I think we can disable compression at 1.5.1, but allow reading of > compressed files for backward compatibility. Why introduce a regression if there is an easy alternative solution (use the 1.4.x code)? > Anyway, this feature can > be replaced by the embedding feature that may be introduced in 1.6.0. But 1.5.1 is not 1.6.0. Georg PS: Somebody should implement an easy to run (on all platforms) test suite. This bug would have been found by it.
