On 2012-12-09 05:05, Ryan Schmidt wrote:
> If clang >= 421 fails to build the port, then the blacklist entry
> should not be removed unless the source gets modified to work around
> the clang bug.
Agreed, but how do you know it's really any later version after 421 when
adding the blacklist entry?
On Dec 8, 2012, at 21:14, Rainer Müller wrote:
> What should I specify in case I don't know yet when this will be fixed?
>
> Assume a port is broken with the current build of clang 421.11.66, but
> it is not clear in which build number this will be fixed or, if the
> developers say something like
On 2012-11-25 17:48, Jeremy Huddleston Sequoia wrote:
> # All clang versions blacklisted
> compiler.blacklist-append clang
>
> # clang < 308 blacklisted:
> compiler.blacklist-append {clang < 308}
>
> # clang >= 421.11.66 and < 444 blacklisted
> compiler.blacklist-append {clang >= 421.11.16 < 444}
On 2012-12-05 22:09, Ryan Schmidt wrote:
> I find that tabular format easier to read than what we currently have.
A table works if we only present the version number. However, I would
prefer to save the whole version information string in case we need to
improve our regular expressions for matchin
On Dec 5, 2012, at 10:04, Rainer Müller wrote:
> On 2012-11-27 11:51, Ryan Schmidt wrote:
>> I've created the compiler_blacklist_versions 1.0 portgroup in r100083. It
>> works in my limited testing. Please try it and let me know what you think.
>
> In order to use this PortGroup, we need informat
On 2012-11-27 11:51, Ryan Schmidt wrote:
> I've created the compiler_blacklist_versions 1.0 portgroup in r100083. It
> works in my limited testing. Please try it and let me know what you think.
In order to use this PortGroup, we need information in which version of
Xcode a specific compiler versi
On Nov 25, 2012, at 20:11, Jeremy Huddleston Sequoia wrote:
> Pseudocode for the parsing is essentially:
> if is-list(l):
>if matches_all_versions(car(l), cdr(l))
>refuse_compiler(l)
> else
>refuse_compiler(l)
>
> matches_all_versions(c, vl):
>if is_empty_list(vl):
>r
On Nov 25, 2012, at 20:11, Jeremy Huddleston Sequoia
wrote:
> for the parsing is essentially:
> if is-list(l):
>if matches_all_versions(car(l), cdr(l))
>refuse_compiler(l)
> else
>refuse_compiler(l)
>
> matches_all_versions(c, vl):
>if is_empty_list(vl):
>return tru
On Nov 25, 2012, at 20:11, Jeremy Huddleston Sequoia
wrote:
> On Nov 25, 2012, at 3:10 PM, Craig Treleaven wrote:
>
> For mythtv-core.25, I tested a bunch of compilers to find a few that worked.
> Eg:
>>
>> # See https://trac.macports.org/ticket/35934
>> compiler.blacklist llvm-gcc-4.2 ma
On Nov 25, 2012, at 3:10 PM, Craig Treleaven wrote:
> At 11:48 AM -0500 11/25/12, Jeremy Huddleston Sequoia wrote:
>> On Nov 25, 2012, at 8:54 AM, Bradley Giesbrecht wrote:
>>
>>> On Nov 24, 2012, at 8:10 PM, Ryan Schmidt wrote:
>>>
One idea that occurs to me is modifying the compiler.bl
At 11:48 AM -0500 11/25/12, Jeremy Huddleston Sequoia wrote:
On Nov 25, 2012, at 8:54 AM, Bradley Giesbrecht wrote:
On Nov 24, 2012, at 8:10 PM, Ryan Schmidt wrote:
One idea that occurs to me is modifying the compiler.blacklist
syntax. For example:
"compiler.blacklist clang" would black
On Nov 25, 2012, at 8:54 AM, Bradley Giesbrecht wrote:
> On Nov 24, 2012, at 8:10 PM, Ryan Schmidt wrote:
>
>> One idea that occurs to me is modifying the compiler.blacklist syntax. For
>> example:
>>
>> "compiler.blacklist clang" would blacklist all versions of clang, as it does
>> now.
>>
On Nov 24, 2012, at 8:10 PM, Ryan Schmidt wrote:
> One idea that occurs to me is modifying the compiler.blacklist syntax. For
> example:
>
> "compiler.blacklist clang" would blacklist all versions of clang, as it does
> now.
>
> "compiler.blacklist clang<318.0.61" would blacklist clang builds
Usually the reason why a port cannot build with clang is because of bugs in the
port's software. In those cases we fix the code if we can and send it upstream;
if we cannot, then we "compiler.blacklist clang". But sometimes it's because of
a bug in clang that gets fixed in a later version of cla
14 matches
Mail list logo
