On 11/11/2015 09:01, Dimitry Sibiryakov wrote:
>Hello, Adriano.
>
>Key holder example can be made more simple and readable by using multiple
> inheritance.
> With normal classes I would do it this way:
>
> class CryptKeyHolder: public IKeyHolderPlugin, public ICryptKeyCallback
>
>Is
11.11.2015 12:32, Adriano dos Santos Fernandes wrote:
> I can write almost any code mixing their common part in a single class,
> avoiding duplicate declarations, with our without cloop, and as the one
> you say, it will not work too, but will be much more compact!
Multiple inheritance of
Hello, Adriano.
Key holder example can be made more simple and readable by using multiple
inheritance.
With normal classes I would do it this way:
class CryptKeyHolder: public IKeyHolderPlugin, public ICryptKeyCallback
Is it possible with CLOOP templates?
--
WBR, SD.
11.11.2015 12:10, Adriano dos Santos Fernandes wrote:
> Why it can be made more simple and readable this way?
No need to use delegation. No second object declaration, initialization and
cross-references with "parent" member. The code is two times shorter.
--
WBR, SD.
On 11/11/2015 09:18, Dimitry Sibiryakov wrote:
> 11.11.2015 12:10, Adriano dos Santos Fernandes wrote:
>> Why it can be made more simple and readable this way?
>No need to use delegation. No second object declaration, initialization
> and
> cross-references with "parent" member. The code is
11.11.2015 01:25, liviusliv...@poczta.onet.pl wrote:
>
> Is this pull request useful at all or this is useless for FB team?
> I try to start to teach FB code, to work in the future on more useful and
> harder topics.
> But i must get start point, will be good to see feedback (and also
> criticism)
On 11/11/2015 08:36, Dmitry Yemanov wrote:
>
> However, choosing XML over JSON or vice versa is IMHO a way nowhere.
> Whatever one we choose, half of users will be against the chosen format
> and ask us to support the second one. So we should either support both
> out of the box or somehow
11.11.2015 12:52, Adriano dos Santos Fernandes wrote:
> Definitively, not always out of the box when they have common base classes.
Even in this case it works predictable in standard C++.
>> >So, what is your answer: is using of multiple inheritance possible with
>> > CLOOP templates
>>
I have been happily using cmake to build fb3 on Linux.
I was therefore disappointed to discover that the cmake files are not
included in th fb3-rc1 tarball.
Would it be possible to include them in RC2, possibly with a warning
note?
Regards, Treeve
On 11/11/2015 03:51 PM, Dimitry Sibiryakov wrote:
> Hello, All.
>
> If there is several encrypted databases on a server, each using the same
> cryto-plugin
> but different keys, how can a key holder find out which key it must provide
> on request?
>
>
I see no problems from key holder -
On 11/11/2015 05:15 PM, Dimitry Sibiryakov wrote:
> Hello, All.
>
> Check me, please: in CryptoManager::encrypt() I see line
>
> > from->pag_flags |= Ods::crypted_page;
>
> But AFAIU, content of "from" page is not encrypted, so if this page is
> used anywhere
> later, an attempt to
On 11/11/2015 11:29 AM, Dimitry Sibiryakov wrote:
> 11.11.2015 16:22, Jim Starkey wrote:
>> Multiple inheritance with common base classes is, in fact, a mess. The
>> problem is that the C++ compiler can't really know whether two like
>> named methods in different classes are the same and has no
11.11.2015 18:03, Jim Starkey wrote:
> Implementing an interface doesn't require
> multiple inheritance -- or any inheritance, for that matter -- just a
> decently designed language.
Yes. But in this case the design requires implementing of two different
interfaces at
once. And inheritance
list(distinct) doesn't sort any more if chars < a are leading
-
Key: CORE-5011
URL: http://tracker.firebirdsql.org/browse/CORE-5011
Project: Firebird Core
Issue Type: Bug
11.11.2015 16:47, Vlad Khorsun wrote:
> I think key name (id) used for encryption should be stored at header page
> as crypto-plugin name (id) does.
Good idea. May be a text UUID shown by gstat can do the trick. It may be
even the same
UUID that is used for nbackup.
--
WBR, SD.
On 11/11/2015 05:22 PM, Dimitry Sibiryakov wrote:
> 11.11.2015 15:15, Alex Peshkoff wrote:
>> Do you want to suggest to provide to plugin info about something like
>> DbName? May be anything else?
> There definitely should be a way for crypto-plugin or a key holder (or,
> better, both)
> to
11.11.2015 16:15, Alex Peshkoff wrote:
> On 11/11/2015 04:54 PM, Dimitry Sibiryakov wrote:
>> 11.11.2015 14:52, Alex Peshkoff wrote:
>>> I see no problems from key holder - key name should be passed to it.
>> Ok.
>> How can crypto-plugin find out which key name it must provide to key
>>
11.11.2015 16:22, Jim Starkey wrote:
> Multiple inheritance with common base classes is, in fact, a mess. The
> problem is that the C++ compiler can't really know whether two like
> named methods in different classes are the same and has no way to known
> which to reference in vtable. This shows
On 11/11/2015 09:57, Dimitry Sibiryakov wrote:
>Right now, in current state of source tree, will work a class derived from
> IKeyHolderPluginImpl and ICryptKeyCallbackImpl? Yes or no?
>
Sorry, but this is something you can verify yourself.
I guess it will not, as even without the magics
11.11.2015 14:52, Alex Peshkoff wrote:
> I see no problems from key holder - key name should be passed to it.
Ok.
How can crypto-plugin find out which key name it must provide to key holder?
--
WBR, SD.
--
Hello, All.
For every shadow a database page is encrypted separately. Why?
--
WBR, SD.
--
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
add missing cmake files to fb3 linux tarball
Key: CORE-5008
URL: http://tracker.firebirdsql.org/browse/CORE-5008
Project: Firebird Core
Issue Type: Improvement
Components: Build Issues
I've downloaded snapshot WI-V3.0.0.32165 and unpacked it.
Then:
c:\> type %fb3_home%\install_service.bat
@echo off
if "%1"=="" (
instreg install -z
)
setlocal
set INSTALL_SERVICE=install -z
set START_SERVICE=start
if not "%1"=="" (
set INSTALL_SERVICE=%INSTALL_SERVICE% -n %1
set
On 11/11/2015 06:22 PM, Dimitry Sibiryakov wrote:
> 11.11.2015 15:21, Alex Peshkoff wrote:
>> That flag means here 'page must be encrypted when written to the disk'.
> Стоп, я опять не понимаю как это не работает.
> Страница читается с диска с установленным флагом. Она расшифровывается,
>
On 11/11/2015 6:52 AM, Adriano dos Santos Fernandes wrote:
> On 11/11/2015 09:37, Dimitry Sibiryakov wrote:
>> 11.11.2015 12:32, Adriano dos Santos Fernandes wrote:
>>> I can write almost any code mixing their common part in a single class,
>>> avoiding duplicate declarations, with our without
11.11.2015 15:21, Alex Peshkoff wrote:
> That flag means here 'page must be encrypted when written to the disk'.
Стоп, я опять не понимаю как это не работает.
Страница читается с диска с установленным флагом. Она расшифровывается, но
флаг
остаётся установленным.
Если идёт процесс
11.11.2015 16:24, Alex Peshkoff wrote:
> What about database name - I think it can be useful not for
> crypt-related plugins only. Therefore I plan to add it to PluginConfig
> interface, both alias and filename. What else worth adding there?
I think that access to current IAttachment* can be
11.11.2015 13:44, Adriano dos Santos Fernandes wrote:
> I guess it will not, as even without the magics behind in the code, it
> would not gonna work directly.
You guess? You don't know how your code work?..
--
WBR, SD.
On 11/11/2015 10:49, Dimitry Sibiryakov wrote:
> 11.11.2015 13:44, Adriano dos Santos Fernandes wrote:
>> I guess it will not, as even without the magics behind in the code, it
>> would not gonna work directly.
>You guess? You don't know how your code work?..
>
Well, I initially would not
11.11.2015 15:15, Alex Peshkoff wrote:
> Do you want to suggest to provide to plugin info about something like
> DbName? May be anything else?
There definitely should be a way for crypto-plugin or a key holder (or,
better, both)
to get some information about database they are loaded for, but
UDF returning 0 for a varchar pointer aborts engine
---
Key: CORE-5007
URL: http://tracker.firebirdsql.org/browse/CORE-5007
Project: Firebird Core
Issue Type: Bug
Components: UDF
11.11.2015 13:55, Adriano dos Santos Fernandes wrote:
> Well, I initially would not gonna to reply to your messages, as I saw
> you were just lazy, just replied because you directed it to me, but you
> were just lazy and wanna I do your job.
Documenting of your design is my job? Wow...
--
On 11/11/2015 09:37, Dimitry Sibiryakov wrote:
> 11.11.2015 12:32, Adriano dos Santos Fernandes wrote:
>> I can write almost any code mixing their common part in a single class,
>> avoiding duplicate declarations, with our without cloop, and as the one
>> you say, it will not work too, but will be
Hello, All.
If there is several encrypted databases on a server, each using the same
cryto-plugin
but different keys, how can a key holder find out which key it must provide on
request?
--
WBR, SD.
--
On 11/11/2015 04:17 PM, Treeve Jelbert wrote:
> I have been happily using cmake to build fb3 on Linux.
> I was therefore disappointed to discover that the cmake files are not
> included in th fb3-rc1 tarball.
>
> Would it be possible to include them in RC2, possibly with a warning
> note?
Please
Hello, All.
Check me, please: in CryptoManager::encrypt() I see line
> from->pag_flags |= Ods::crypted_page;
But AFAIU, content of "from" page is not encrypted, so if this page is used
anywhere
later, an attempt to decrypt unencrypted data will result in garbage.
Am I wrong?
--
On 11/11/2015 04:54 PM, Dimitry Sibiryakov wrote:
> 11.11.2015 14:52, Alex Peshkoff wrote:
>> I see no problems from key holder - key name should be passed to it.
> Ok.
> How can crypto-plugin find out which key name it must provide to key
> holder?
>
Using plugin's configuration this
> Generally, providing some structural information in the machine-readable
> form is a good idea. Parsing plain text is terrible, especially when
> this text changes from version to version.
>
> When I was implementing the explained plan output, the original idea was
> to have it represented
>>Try to write code similar to the one you're changing.
good point :)
regards,
Karol Bieniaszewski
W dniu 2015-11-11 02:40:05 użytkownik Adriano dos Santos Fernandes
napisał:
> About the code style...
>
> You surely should not reinvent another style. It's very easy to
Hi,
great thanks - this is detailed information
regards,
Karol Bieniaszewski
W dniu 2015-11-11 06:25:08 użytkownik Dmitry Yemanov
napisał:
> 11.11.2015 04:40, Adriano dos Santos Fernandes wrote:
>
> > About the code style...
> >
> > You surely should not reinvent another
40 matches
Mail list logo