Re: How to get Ada to ?cross the chasm??

["Randy Brukardt" <randy@rrsoftware.com>]

"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote in message 
news:pd0ukp$7mu$1@gioia.aioe.org...
> On 2018-05-09 23:33, Paul Rubin wrote:
>> "Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> writes:
...
>>> It is incredibly difficult to estimate the time spent by the clients
>>> of GC. What is measured is probably the GC task with all distributed
>>> overhead in the clients and the overhead of the overall design
>>> uncounted.
>>
>> Hmm, this is an interesting claim that I won't say is wrong, but I
>> haven't heard it before and am surprised by it and am skeptical.

At least in a language like Ada, reachability tracking is itself a 
complicated problem. One needs some sort of code thunk or data structure to 
be able to find all of the stand-alone access objects/parameters (these are 
the ones that root data structures), including the ones buried in private 
types. The overhead of such a structure/code is clearly outside of the GC 
modules themselves (it has to belong to the compiled user code, so it 
doesn't get picked up by standard profiling). Whether that overhead is 
mainly in time or mainly in spaace can be traded off, but that won't change 
for a particular implementation (implementing that once surely is enough).

Comparing that to an environment that doesn't use GC is always going to be 
difficult. On top of which, one really ought to include any extra overhead 
incurred in explicit storage management (finalization, for instance) if one 
was going to get a really fair answer. This is a tough problem, any real 
answer would only be valid for a particular pair of programs.

I personally prefer to have my complications explicit, so I can attack them 
if performance improvement is needed. YMMV.

                               Randy.