Initializing structures for callstack processing (building memory map of the
process, gathering kernel symbols, etc) takes some time, which in some cases
may be significant.
Callstack queries are now handled on a separate thread. In such setup it no
longer makes sense to block main thread execution with this lengthy init
process.
All the heavy initialization phase has been now moved to this separate
processing thread. Some initial callstack queries may now not produce
responses as promptly as before, but this is only because the main thread is
able to start working earlier.
Some parts of the initialization process may be critical to do in the main
thread, for example because the function responsible for gathering callstacks
must be loaded first. This is done still on the main thread, in a new function
InitCallstackCritical().
Note that the negative glyph spacing is const here, regardless of the font
scaling. The alternative of using proportional spacing in reality does not
work as good as it sounds. With fractional scaling some of the glyphs will
no longer land on exact pixel boundaries, which results in blurry text.
Looking at the potential solutions for this problem is moot point, as with
the non-fractional scale of 2x and glyph spacing of -2, there is no longer
enough space separating some letters, which is especially visible in words
like "common" or register "xmm", where the "mm" pair becomes joined.
"Always full" was kinda ambiguous. It could either mean that the name will be
always full (shortening effectively disabled), or that the shortening will be
performed to the maximum extent (as it was).
Namespace shortening was kinda ok for function names produced by MSVC, which
are generally clean looking. However, gcc/clang like to produce function names
which include template arguments, function parameters, return values, etc. In
such cases the old algorithm simply didn't work, because removal of everything
before the last :: could as well happen in midst of function parameters list.
The result was certainly not an usable function name.
With this new approach namespaces are no longer explicitly mentioned and this
functionality is simply called zone name shortening.
The user-selectable options were changed to make the shortening always
enabled, disabled, or to apply as needed. Note that the "as needed" approach
will be dynamic, trying to gradually remove more and more from the name, until
it fits in the requested area.
Current implementation is only the first step into making this work. In this
first step the function parameters are reduced to () and the template
arguments are reduced to <>. This alone greatly improves readability of the
zone names.
The option to reduce namespaces to one letter (i.e. std::tr1::hash would
become s:t:hash) will no longer be present, now or in the future.
Bartosz Taudul