Re: [boost] [interprocess] Just a note of thanks to the developers:)

Subject: Re: [boost] [interprocess] Just a note of thanks to the developers:)
From: Faisal Vali (faisalv_at_[hidden])
Date: 2008-09-06 12:57:48

"Ion Gaztañaga" <igaztanaga_at_[hidden]> wrote in message 
news:48C1AACA.5090305_at_gmail.com...
> Faisal Vali wrote:
>>
>> Well, I was just wondering what was the reason for not including a 
>> recursive_named_upgradeable_mutex?
>
> I just didn't see any use case for that. Is there any C++ synchronization 
> library that offers this kind of mutex? Why do you need it?
>

I had created wrapper functions for my containers that were synchronized 
using a named upgradeable mutex and lock guards, and ended up having to call 
one function from the other (which causes deadlock if an exclusive lock is 
got) - i thought a recursive lock would have allowed this (not sure the pros 
and cons of this approach).  But since there is no such mutex, i settled on 
a couple of solutions: one involved creating a scoped_lock appropriately if 
a bool param indicates a lock-recusrive call or not, the other uses 
templatized versions that use an mpl::if_c to create noop locks if the 
nontype bool param indicates to do so.

Please keep in mind that I am still learning about synchronization issues 
(and am light on the theory, so i might have missed a common pattern for 
solving this type of problem)
thanks,
-fas