Summary says it all, except that this fails miserably with gcc 3.4, where the following runhugs errors occur: 1) Syntax error in declaration (unexpected `{', possibly due to bad layout), for a lot (if not all) of the files in ${S}/hugsdir/libraries/Data (e.g. Bits.hs, Lists,hs, Ratio.hs). 2) Last generator in do {...} must be an expression, for ${S}/hugsdir/libraries/System/IO/Error.hs 3) Unterminated nested comment {- ..., for ${S}hugsdir/libraries/Foreign/C/Types.hs 1) is almost certainly a whitespace problem, and I wouldn't be suprised if 2) and 3), but I haven't had time to look at them yet Reproducible: Always Steps to Reproduce:
Created attachment 43684 [details] hugs98-20041101.ebuild
hugs does not seem to be gcc 3.4 friendly yet, it's a C pre-processor problem. See bug #64426 for more details and a possible patch/solution. We'll add an ebuild for hugs when the next release is made (which according to the latest HC&A should be soon). We should get our gcc 3.4 patches upstream to the hugs maintainers so we don't have to keep applying the patches in the ebuild.
I came up with the fix for bugs #64426, but sadly this is a different problem. Upstream already have a fix for #bug 64426, which is slightly different - instead of changing "infix 5 \\" in whatever file it is to "infix 5 \\\", they changed it to "infix 5 \\ -- comment to fool cpp". The end result is the same either way though.
I agree that we should make sure that whenever hugs98 will be released again, it will be gcc-3.4-friendly. Apart from testing this now, is there anything in the current hugs98 development snapshot that you need and that isn't in Nov 2003? ks
No, there's nothing in it that I need, I just thought that any one feeling adventurous might like to help test this, and it gives us a head start for when the next stable release is made
I think this can be closed. Afaics, the Hugs interim release builds fine with gcc-3.4. ks