On 03/15/2016 12:40 PM, Dimitry Sibiryakov wrote:
> 15.03.2016 10:33, Alex Peshkoff wrote:
>> Where do we have races in_current_ codebase?
> Dunno, you told about them.
>
I've said that _if_ you remove filenames comparison from
initAttachment() in jrd.cpp or make it case-insensitive on posix
15.03.2016 10:33, Alex Peshkoff wrote:
> Where do we have races in_current_ codebase?
Dunno, you told about them.
--
WBR, SD.
--
Transform Data into Opportunity.
Accelerate data analysis in your applications with
On 03/14/2016 08:56 PM, Dimitry Sibiryakov wrote:
> 14.03.2016 18:39, Alex Peshkoff wrote:
>> Do you suggest to lock db list mutex from yvalve? And from network
>> server? :)
> This is a separate discussion about thread races on database file
> creation and open, if
> you haven't noticed. Y-va
14.03.2016 22:51, Vlad Khorsun wrote:
> You must consider detach, drop and database shutdown to get full picture.
Yes, and all the same for classic mode as well, so the picture is quite
complex.
But MSDN promises file ID to be stable as long as at least one handle is open,
so there
may b
14.03.2016 19:29, Dimitry Sibiryakov wrote:
> 14.03.2016 18:12, Alex Peshkoff wrote:
>> But use of by name search if ID is unusable due to file not
>> exists appears unavoidable.
>
> Not true. Thread that is going to create file must do it while keeps db
> list mutex. In
> this case it can be
> 14.03.2016 20:25, Leyne, Sean wrote:
> > - try to open/create file with non-shared access attribute, if success then
> get fileID else fail/ throw "file already exists"
>
>In superserver at this point other thread can attach to this file and get
> broken (non-initialized) database.
Actual
14.03.2016 20:25, Leyne, Sean wrote:
> - try to open/create file with non-shared access attribute, if success then
> get fileID else fail/ throw "file already exists"
In superserver at this point other thread can attach to this file and get
broken
(non-initialized) database.
> - create dbb
>I.e. sequence like this:
>
> 1) create dbb with flag "creating"
> 2) get list mutex
> 3) create the file and get its id
1- This would destroy any existing file, no?
2- why "get list mutex" before the create file and/or get fileID?
Wouldn't the better "creating" order be:
- try to open/cr
14.03.2016 18:39, Alex Peshkoff wrote:
> Do you suggest to lock db list mutex from yvalve? And from network
> server? :)
This is a separate discussion about thread races on database file creation
and open, if
you haven't noticed. Y-valve has nothing to do with that.
--
WBR, SD.
On 03/14/2016 08:29 PM, Dimitry Sibiryakov wrote:
> 14.03.2016 18:12, Alex Peshkoff wrote:
>> But use of by name search if ID is unusable due to file not
>> exists appears unavoidable.
> Not true. Thread that is going to create file must do it while keeps db
> list mutex.
Do you suggest to lo
14.03.2016 18:12, Alex Peshkoff wrote:
> But use of by name search if ID is unusable due to file not
> exists appears unavoidable.
Not true. Thread that is going to create file must do it while keeps db list
mutex. In
this case it can be sure that no other thread can try to create the file in
11 matches
Mail list logo
