diff options
Diffstat (limited to 'include/tools/long.hxx')
| -rw-r--r-- | include/tools/long.hxx | 15 |
1 files changed, 9 insertions, 6 deletions
diff --git a/include/tools/long.hxx b/include/tools/long.hxx index ef707070af8b..5e0800f2c274 100644 --- a/include/tools/long.hxx +++ b/include/tools/long.hxx @@ -13,13 +13,16 @@ /** * We have an unfortunate problem in that, on Windows (both 32 and 64-bit), long is always 32-bit. * But on Linux (64-bit) long is 64-bit. Which leads to nasty situations where things that work - * on Linux, fail on Windows. + * on Linux, failed on Windows. * So....., - * (1) introduce a typedef that will replace all usage of long in our codebase. - * (2) to start with, that typedef acts that same as the long it replaces - * (3) once the replacement is complete, this typedef will, on windows, become 64-bit. - * (4) and once the fallout from (3) is fixed, we can start replacing tools::Long with things - * like sal_Int64. + * (1) introduce a typedef that replaces (almost) all previous usage + * (2) on 64-bit Windows, this typedef is forced to 64-bit (32-bit platforms, including 32-bit Windows, are left alone) + * (3) fix fallout from (2) + * + * As a consequence of the above, it would be best, over the long term, to regard usage + * of tools::Long in the codebase as meaning "we're not sure what the ideal size of the datatype is", + * and where possible, replace it with a better datatype like sal_Int32/sal_Int64/etc. + * */ namespace tools { |