> 21.06.2013 10:23, Vlad Khorsun wrote:
>> On client side only
>
> I don't see it in inet.cpp:aux_connect(). Is it somewhere else?
Its there. Near line 1540 (B2_5_Release):
address.sin_family = AF_INET;
address.sin_port = ((struct sockaddr_in
*)(response->p
> Again, I ran into the problem, that under certain conditions names can
> not be resolved in (recursive) CTEs. I'm pretty sure it's the same
> problem as described in http://tracker.firebirdsql.org/browse/CORE-4005
> - thus I'm not going to create a new issue.
Feel free to add your case into
>> Feel free to add your case into the ticket, it could be useful when issue
>> will be fixed and tested.
>
> Thank you, Vlad, but I fear, that one is the most reduced testcase I can
> offer (actually, Frank, who posted the issue, is a colleague of mine) -
> the "real live"-examples are quite
> I'd like to select from a FB2.5-DB with FB3.0 and get the error
> "username and password are not defined" (they sure are, I can run the
> statement from another FB2.5-database), and I assume, the problem is the
> encrypted password.
>
> Is there a parameter for "execute statement" to tell that "
> I just tried and get the message "unsupported on-disk structure; found
> 11.2, support 12.0" :-( ...
>
> Is there a way to get around that?
Because you attaching to the FB3 :) Tell us more about your setup:
1. host name and TCP port for fb3 instance
2. host name and TCP port for fb25 instan
- Original Message -
From: "Adriano dos Santos Fernandes"
To: "firebird-devel"
Sent: Sunday, August 04, 2013 6:25 AM
Subject: [Firebird-devel] Firebird 3 crash - Timer
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7577b700 (LWP 18276)]
0x6
Adriano,
any details ? is it reproducible ?
Regards,
Vlad
--
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you t
> I've got dump files available; if someone in particular would like them, I'll
> zip and provide a download URL, privately.
Provide it to me, please. But i'm afraid it is impossible to fix it without
reproducing :(
Regards,
Vlad
On 4-8-2013 15:21, Mark Rotteveel wrote:
...
Interestingly, if I do a trace, I get the following:
Statement 145:
---
SELECT RDB$DB_KEY,"ID","LONG_STR","STR","CamelStr"
FROM TEST_TABLE2 WHERE
RDB$DB_KEY = ?
param0 = var
> Configuration fbtrace.conf contains syntax errors
> -
>
> Key: CORE-4163
> URL: http://tracker.firebirdsql.org/browse/CORE-4163
> Project: Firebird Core
> Issue Type: Bug
> Components: E
> On 09/04/13 11:53, Dmitry Yemanov wrote:
>> 04.09.2013 11:43, Alex Peshkoff wrote:
>>> May be in that case next step? Keep declaring strings as char(N), but
>>> treat N not as maximum size of string
>> Please don't play against the SQL standard. The declared string length
>> is a constraint which
>> > In tracker case 4212, Vlad Khorsun wrote:
>> >
>> > It is documented since introducing GTT in v2.1 that
>> > constraints between temporary and permanent tables is forbidden. This
>> requirement is per SQL standard.
>> >
>> >
> In tracker case 4212, Vlad Khorsun wrote:
>
> It is documented since introducing GTT in v2.1 that constraints
> between temporary and permanent tables is forbidden. This
> requirement is per SQL standard.
>
> I want to be sure that this is correct (sorry Vlad
> -- Original Message ---
> From: "Vlad Khorsun"
>> It is physically impossible to check detail data in all instances
>> of detail GTT when master record is changed.
>>
>> Regards,
>> Vlad
> --- End of Original Message ---
> Hi!
>
> Debugging CORE-4213, I found that ALTER DOMAIN uses the default database
> charset.
>
> -- Database default charset is WIN1252.
>
> -- OK: Domain is created with WIN1252.
> create domain d as varchar(10);
>
> -- OK: Charset of domain is changed.
> alter domain d type varchar(10) chara
> I seem to have found it in jrd/inf.cpp in Firebird 3, it could use the
> vax int64 format, but the actual update counts are still an unsigned 32
> bit integer.
>
> Are there any plans for supporting longer update counts?
Add request for it at tracker, please.
Regards,
Vlad
-
> Any suggestions, guys? Can anyone please point us to the right place in the
> source code so we can have a closer look?
Physical layout of unpacked record in memory is simple:
- bitmak of NULL values, one bit per field
This bitmask contains one 32-bit intereg per every 32 fields in rel
> Specify the name of the table, if an exception occurs when insert
> -
>
> Key: CORE-4252
> URL: http://tracker.firebirdsql.org/browse/CORE-4252
> Project: Firebird Core
> Issue Typ
> On Fri, 25 Oct 2013 16:31:56 +0200, Dimitry Sibiryakov
> wrote:
>> 25.10.2013 16:17, Vlad Khorsun wrote:
>>> Also, i consider accompanied SQL CODE message
>>>
>>> The insert failed because a column definition includes
>>> va
>>> BTW: Would it be possible to include the name of the actual constraint
>>> that was violated?
>>
>> CREATE TABE T
>> (
>> F1 INT NOT NULL,
>> F2 INT CHECK(F2 >= 0)
>> )
>>
>> what constraint name would you like to report in both cases ? :)
>>
>> Regards,
>> Vlad
>>
>
> In the first case
>>> BTW: Would it be possible to include the name of the actual constraint
>>> that was violated?
>>
>> CREATE TABE T
>> (
>> F1 INT NOT NULL,
>> F2 INT CHECK(F2 >= 0)
>> )
>>
>> what constraint name would you like to report in both cases ? :)
>
> Doesn't that get assigned an internal name (
> Final test took 26 minutes, it looks like the majority of the waiting
> time is in the creation of a database (almost every individual test
> creates a database), which takes 4-5 seconds, instead of < 1 second per
> database.
Is it reproducible with isql ?
Regards,
Vlad
> On 3-11-2013 18:14, Vlad Khorsun wrote:
>>> Final test took 26 minutes, it looks like the majority of the waiting
>>> time is in the creation of a database (almost every individual test
>>> creates a database), which takes 4-5 seconds, instead of < 1 second p
> 03.11.2013 18:52, Vlad Khorsun wrote:
>>> Yes, isql shows the same behavior.
>> I can't reproduce it with just build fb3
>
> I can.
> Just unpacked today's snapshot, run 'fbserver -a',
There is no "fbserver" in fb3.
&
> 03.11.2013 20:03, Vlad Khorsun wrote:
>> "create database" runs instantly for me.
>
> For me - not.
> You can get Process Monitor's log here:
> http://www.ibphoenix.com/ibpr_devel/Logfile.7z
> It seems that most of the time were spend in loo
>> PS Adriano already complains about slowness of CREATE DATABASE in fb3
> (see
>> thread
>> "Reason for slow CREATE DATABASE in FB 3 (Linux)" started at 15 dec
>> 2011), i'll look
>> at it again in a next few days.
>
> As Alpha 1 doesn't show this performance problem, it is either a
> re
>> PS Adriano already complains about slowness of CREATE DATABASE in fb3 (see
>> thread
>> "Reason for slow CREATE DATABASE in FB 3 (Linux)" started at 15 dec
>> 2011), i'll look
>> at it again in a next few days.
>>
>> As far as I remember, at least one issue was related with changed
>> de
> I have a desktop, database is written to a RAID-0 (striped).
Could you made events trace with ProcessMonitor ?
Regards,
Vlad
--
Android is increasing in popularity, but the open development platform that
developers
> On 4-11-2013 11:09, Vlad Khorsun wrote:
>>> I have a desktop, database is written to a RAID-0 (striped).
>>
>> Could you made events trace with ProcessMonitor ?
>
> https://www.dropbox.com/s/gzwppacaya1qyme/ProcMon_FB_CreateDatabase.7z
>
> Is this s
> Hello, what's the performace penalty for using an AtomicCounter?
It is noticeable (comparing with single int's) but common effect depends on
how often it
is used. In short - it could be measured in CPU-bound load and it is close to
zero at
IO-bound load.
> I see in
> cch.h, inside
> cl
>> This change seems to be absolutely safe no matter of what more
>> performance optimizations we will try later.
>
> I'd also start with this solution first.
Done.
Regards,
Vlad
--
November Webinars for C, C++, For
> On 04/11/2013 07:52, Vlad Khorsun wrote:
>>>> PS Adriano already complains about slowness of CREATE DATABASE in fb3 (see
>>>> thread
>>>> "Reason for slow CREATE DATABASE in FB 3 (Linux)" started at 15 dec
>>>> 2011), i'
> On 7-11-2013 11:34, Vlad Khorsun wrote:
>
>> Fix is committed, please verify.
>
> Jaybird testsuite with 3.0.0.70717 runs in 4 minutes 37 seconds instead
> of the 26+ minutes with 3.0.0.30708.
Looks better but i don't know how it was with alpha1 a
> On 07-11-2013 08:34, Vlad Khorsun wrote:
>>
>> Fix is committed, please verify.
>>
>
> It's better than it was, but 3.0 is still much slower than 2.5
> fb_smp_server. Look at TCS times:
Probably it is because in v3 FW is OFF until create_database c
> On 10/11/2013 12:43, Vlad Khorsun wrote:
>>> On 07-11-2013 08:34, Vlad Khorsun wrote:
>>>> Fix is committed, please verify.
>>>>
>>> It's better than it was, but 3.0 is still much slower than 2.5
>>> fb_smp_server. Look at TCS
> On 11/11/13 15:25, Vlad Khorsun wrote:
>
>> If you really need to run TSC as fast as with v2.x - set FW=OFF in dpb
>> when
>> database is created.
>>
>
> Taking into an account that when that tests were written FW=OFF was
> default this seems t
> Hi, seems that Commit rev 58928 about CORE-4297 lead to fail build for
Fix is committed. Could you verify, please ?
Vlad
--
Rapidly troubleshoot problems before they affect your business. Most IT
organizations do
> On 12/21/13 12:39, Dmitry Kovalenko wrote:
>> Guten morgen.
>>
>> I have the problem on the client side [32-bit process] with fbclient.dll
>> [v3.0.0.30761, debug build].
>>
>> interface.cpp line 3232
>> 3232| fb_assert(statement->haveException() == 0); //<--
>> 3233| statement->clearExceptio
Hi, all
I will speak about how pages are allocated and released in Firebird. The
current algorithm is well
known, simple and easy to both understanding and implementation:
- we have single bitmap in database where every bit corresponds to one page,
- this bitmap is stored at sequence of P
> I'm not sure this will change much. It will reduce the number of times
> that the PIP needs to be referenced, which will save a little CPU.
Also it will lower concurrency on PIP pages. Of course, common level of
concurrency will not became much lower, but anyway...
> Careful write need
>> I can provide patch or compiled binaries (for Windows) for testing to
>> interested.
>
> I would not give up the compiled binary (for Windows x86). Very
> interesting to compare the performance of cold filling cache.
I'll prepare binaries at this weekend.
Thanks,
Vlad
---
> Sounds good.
Thanks
> Have you thought about tablespaces as the next step?
Not now :)
Regards,
Vlad
--
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a
>> I would not give up the compiled binary (for Windows x86). Very
>> interesting to compare the performance of cold filling cache.
>
>I'll prepare binaries at this weekend.
Here they are:
http://web.firebirdsql.org/downloads/rabbits/hvlad/Firebird-3.0.0.30811_Win32-Extents.7z
http://w
> I’ve got build 30793 of Firebird 3, zip file. Unzipped it,
> started with –a –p as an application.
By default, shared database access is ON (i.e. SS mode in the engine is ON).
Without switch -m you started listener which will launch processes for every
attachment
(i.e. CS mode in the remote
>>> I would not give up the compiled binary (for Windows x86). Very
>>> interesting to compare the performance of cold filling cache.
>>
>>I'll prepare binaries at this weekend.
>
>Here they are:
>
> http://web.firebirdsql.org/downloads/rabbits/hvlad/Firebird-3.0.0.30811_Win32-Extents.
> The reason I was trying this was to try and answer Pavel's question - how do
> you run SuperClassic under FB3 on Windows? AFAICT it is not possible.
Really ? :)
> I don't mind - SuperClassic seems to share the worst features of SuperServer
> and
> ClassicServer, and we should officiall
>> Just a wild idea.
>>
>> We introduce a separate (very thin) fblistener.exe (fbremote or
>> whatever) binary that acts similar to xinetd on posix. It will spawn
>> firebird.exe per every user connection. firebird.exe will detect this
>> situation and explicitly pass SharedCache = false and Shared
>> Technically it is possible to make single instance of firebird.exe to handle
>> some attachment requests by forking new CS process and another attachment
>> requests handling by itself (SS\SC).
>
> Do you mean forking the worker after op_connect / op_attach are
> processed in the "multi-server
> Hello,
>
> the following is with: Firebird-2.5.2.26540-0_Win32 Embedded
>
> What can be the reason for the following error:
>
> "cannot initialize UNICODE collation to use in trace plugin"
Adriano could know, i hope
> when a trace session is attempted to be started with Firebird Embedde
> It is interesting now, that the trace works for an isql based client
> session, even when using UTF8 as character set, but fails with e.g.
> Database Workbench. The only difference is that isql is in the same
> directory as embedded, but DBW isn't, although pointing to the same
> embedded lib
>>> It is interesting now, that the trace works for an isql based client
>>> session, even when using UTF8 as character set, but fails with e.g.
>>> Database Workbench. The only difference is that isql is in the same
>>> directory as embedded, but DBW isn't, although pointing to the same
>>> embedd
> Any further ideas?
Not at me.
Regards,
Vlad
PS If you provide me with reproducible sample i'll try to investigate it
--
WatchGuard Dimension instantly turns raw network data into actionable
security intelligence
>> PS If you provide me with reproducible sample i'll try to investigate it
>
> The only thing I could provide is a 32-bit Delphi application which
> points to an Embedded library. The executable must not be in the same
> directory. Would this be useful?
At least, we can try :)
Regards,
Vl
PS If you provide me with reproducible sample i'll try to investigate it
>>>
>>> The only thing I could provide is a 32-bit Delphi application which
>>> points to an Embedded library. The executable must not be in the same
>>> directory. Would this be useful?
>>
>> At least, we can try :)
>>> You now should see the UNICODE initialization problem in the trace output.
>>
>> No, i see ATTACH_DATABASE event by this application.
>
> That's a pity. ;-)
>
> Do you have the FIREBIRD environment variable set?
No.
>> PS does you tried LoadLibraryEx with LOAD_WITH_ALTERED_SEARCH_P
>>> PS You could also try to use FileMon\ProcMon to trace application activity
>>
>> I see that icuin30.dll is tried to be loaded from various places
>> (although excluding the embedded root path), but this fails.
>>
>> icuin30.dll is available in the embedded root directory. When I put the
>> dll
> 02.02.2014 20:16, Vlad Khorsun wrote:
>> It finally points me to the source of the problem - fbtrace.dll have no
>> idea that it works
>> with embedded and load icuin30.dll not using LOAD_WITH_ALTERED_SEARCH_PATH
>> flag.
>
> I wonder why it doesn'
PS You could also try to use FileMon\ProcMon to trace application activity
>>>
>>> I see that icuin30.dll is tried to be loaded from various places
>>> (although excluding the embedded root path), but this fails.
>>>
>>> icuin30.dll is available in the embedded root directory. When I put the
>
> 12.02.2014 10:11, Vlad Khorsun wrote:
>> PPS I have no idea if this issue is present on non-Windows platforms
>
> This issue is not present on non-Windows platform because ICU is considered
> a system
> library there and thus always is placed in well-known system librar
> 12.02.2014 10:48, Vlad Khorsun wrote:
>> It is not about ICU only. It is about loading libraries using just file
>> name, without a path.
>
> Unix guys aren't afraid of hell and put all libraries or at least symlinks
> to
> /usr/lib/. On the oth
> My application is my application.
So, you have full control over it. Very well.
> Firebird client is a part of customer's environment.
Probably. But it is easy to isolate your application from customer's
environment
and make it use your private copy of fbclient.dll + firebird.msg
> 12.02.2014 23:03, Vlad Khorsun wrote:
>> Probably. But it is easy to isolate your application from customer's
>> environment
>> and make it use your private copy of fbclient.dll + firebird.msg.
>
> Yes. And as the result to get "private environm
> Hello,
>
> when I execute a stored procedure which is part of a package with e.g.:
>
> select * from pkg_schema.s_user_tables;
>
> I guess, it would make sense to also have the package name somewhere in
> the trace output for the EXECUTE_PROCEDURE_* event. Does this make sense?
Good poin
> 13.02.2014 16:51, Dmitry Yemanov wrote:
>> Do you mean really unreadable or actually readable but prefixed with
>> "cannot locate firebird.msg"? Example maybe?
>
> Actually, half-readable. There may be meaningfil lines as well as prefixed
> with "can't
> format message". Example:
>
> unknow
> 13.02.2014 9:19, Vlad Khorsun wrote:
>> Please, provide real sample of the task you can't solve using existing
>> facilities and we will
>> think how to improve it.
>
> As I wrote in ticket: fbclient.dll installed into system directory produces
> unre
>>> Good point. I think it makes sense. I'll look at it (a bit later). Would
>>> you like to
>>> see procedure name prefixed with package name (PACKAGE.PROCEDURE)
>>
>> This would be consistent with the rest of the codebase, so IMHO it
>> should be done this way.
> Sounds good.
Check next sna
13.02.2014 22:02, Vlad Khorsun wrote:
This error code (isc_bad_epb_form) was intorduced in FB 2.5 and it means
you are
used older fbclient.dll with your application when obtained this error from
Firebrid server.
Also it means that firebird.msg corresponding to your old client library also
> 14.02.2014 22:50, Thomas Steinmaurer wrote:
>
>> Would like to do so, but the latest snapshot build for Windows available
>> is from 2 days ago. Just letting you know.
>
> Windows builds are broken at the moment, please wait a bit.
It works now
Regards,
Vlad
-
> will there be or is there already support for stored function execution
> in the Trace API similar to stored procedures?
Sure, it is under development.
Regards,
Vlad
--
Android apps run on BlackBerry 10
Introduci
> Tried todays snapshot build (Win 32-bit). The output looks like a bit
> weird regarding package/procedure name:
>
> 2014-02-18T21:59:55.1080 (456:00C32D20) EXECUTE_PROCEDURE_FINISH
> tourism.fdb (ATT_332, TOURISM:NONE, UNICODE_FSS, TCPv4:127.0.0.1)
> C:\Program Files (x86)\Upscene Produ
>>> will there be or is there already support for stored function execution
>>> in the Trace API similar to stored procedures?
>>
>> Sure, it is under development.
>
> Thanks for the information.
Done, happy testing ;)
Regards,
Vlad
-
>>> Tried todays snapshot build (Win 32-bit). The output looks like a bit
>>> weird regarding package/procedure name:
>>>
>>> 2014-02-18T21:59:55.1080 (456:00C32D20) EXECUTE_PROCEDURE_FINISH
>>> tourism.fdb (ATT_332, TOURISM:NONE, UNICODE_FSS, TCPv4:127.0.0.1)
>>> C:\Program Files (x86)
> I fixed the most obvious bugs in v3 (the ones I don't mention here), but I
> have three tough ones:
>
> 1.- In CCH_down_grade_dbb() I need to know what's the scope for the lock
It was introduced by me and i'll fix it
Regards,
Vlad
--
> One question though ...
>
> When I execute a stored procedure (with or without package, doesn't
> matter), e.g.:
>
> select * from pkg_schema.s_user_tables;
>
> I also see appropriate PREPARE_STATEMENT etc. events for the SELECT
> statement. Whereas when I execute a stored function with e.g.
Fixed.
Regards,
Vlad
--
Flow-based real-time traffic analytics software. Cisco certified tool.
Monitor traffic, SLAs, QoS, Medianet, WAAS etc. with NetFlow Analyzer
Customize your own dashboards, set traffic alert
> On 03/05/2014 05:55 AM, Adriano dos Santos Fernandes wrote:
>> Accordingly to how new API works, where one can get a version and the
>> virtual table is filled with possible non-existent methods, classes like
>> TraceParamsImpl could not insert new virtual functions.
>>
> Yes, and it was clear si
>> On 03/05/2014 05:55 AM, Adriano dos Santos Fernandes wrote:
>>> Accordingly to how new API works, where one can get a version and the
>>> virtual table is filled with possible non-existent methods, classes like
>>> TraceParamsImpl could not insert new virtual functions.
>>>
>> Yes, and it was cl
> What's the purpose for following code in subj?
If you really interested, you can look at few previous revisions of this
code.
>> InternalMessageBuffer inMsg(m_inMetadata, m_in_buffer.begin());
>> InternalMessageBuffer outMsg(m_outMetadata, m_out_buffer.begin());
>>
>> m_request->execute
>I needed comments from developers also
Then post links to working sites ;) I can't access it whole yesterday
and today nothing changed.
Regards,
Vlad
--
Learn Graph Databases - Download FREE O'Reilly Book
"Graph Da
Marius,
> you can check the google cache
Why should i spend my time to investigate (!) how to access it ???
Does it have something important for Firebird ? Something new ?
Or what ?
Regards,
Vlad
--
Learn Grap
> On 03/11/2014 12:22 PM, Dmitry Yemanov wrote:
>> 11.03.2014 09:59, marius adrian popa wrote:
>>
>>> I needed comments from developers also
>> You'd better ask questions then. I don't care to comment what unknown
>> people write about Firebird over the internet.
> +1
++
Vlad
---
>I posted here so maybe it can help us to update the sql conformance
> page http://www.firebirdsql.org/en/sql-conformance/
>
> Sorry for cross posting so much , maybe i was over to much coffeine
> also i didn't had the time to explain what is with the link
>
> ps: he is not quite a random guy on
> I have just started testing FB 3.0 Alpha 2 (32 bits) and I'm getting in
> trouble using the same UDF's I already use in FB 2.5. I'm getting the known
> message:
> "Invalid token.
> invalid request BLR at offset 35.
> function ? is not defined.
> module name or entrypoint could not be found."
> Provide name of TABLE / VIEW when some statement violates CHECK.
>
>
> Key: CORE-4368
> URL: http://tracker.firebirdsql.org/browse/CORE-4368
> Project: Firebird Core
> Issue Type
> Check me, please: at the beginning on VIO_verb_cleanup() pointer to one
> savepoint is
> saved to variable sav_point and excluded from both transaction's used and
> free savepoints
> chains. If any exception is thrown from inside of try-catch block, this
> savepoint is
> leaked. Am I wron
> A definite sequence of 'release savepoint' + 'rollback to savepoint'
> can lead to missing entries.
And you have a sample, of course ?
Regards,
Vlad
--
Firebird-Devel mailing list, web interface at
https://lists.
> 28.03.2014 17:43, Dimitry Sibiryakov wrote:
>>Either I don't understand something or garbage_collect_idx() doesn't do
>> what it is
>> supposed to do.
>
> I've found the problem. Global variables must die. Only miracle prevents
> this piece of
> code from crash:
>
> > undo = action->vc
>> Record buffer is cached at the transaction level and exists in single
>> instance. It is OK
>> while we use just one instance of record buffer at the same time.
>
> What is meant here by "record buffer" -- tra_undo_record or what?
tra_undo_record, of course
Regards,
Vlad
---
> > b) VIO_verb_cleanup() also can't use more than one instance
> > of record buffer itself but it could call verb_post()
> > while record buffer is already used:
>
>
> VIO_verb_cleanup() can call itself recursively, but it seems that because it
> happens at the end of the function, it isn'
> Consider a large database (200GB, large tables with 450 Million rows) which
> is running on a kick-a** server with pool of
> enterprise SSDs for storage (more IOPS then Firebird could ever use), which I
> need to extract data from on a regular basis
> throughout the day for use by an external
>> Instead, i think, we could implement something like "clone transaction"
>> which will:
>> - get "master" transaction number at input
>
> One detail should be taken into an account here - what if user passes
> incorrect transaction number on input.
Of course there should be error rai
> On 04/02/14 15:20, Vlad Khorsun wrote:
>
>>>> Note, it have no sence for read-committed transactions (as they have
>>>> no snapshot to share\clone) and for
>>>> "consistency write" transactions (as they will conflict on relations loc
>> > As far as I can tell, there is no way for this to be done.
>>
>> I'd said you need "common snapshot". I.e. few transactions should use the
>> same snapshot view of database.
>
> Correct.
>
>
>> Instead, i think, we could implement something like "clone transaction"
>
> "Clone Tran
> On 4/2/2014 5:28 AM, Vlad Khorsun wrote:
>>> C
>>> I think it will be not easy to implement from the Y-valve point of
>>> view - i see no simple way to pass handles from one
>>> process
>>> (coordinator) to another (worker)
> Fernandes (JIRA)" wrote:
>> Incorrect line/column information in runtime errors
>> ---
>>
>> Key: CORE-4381
>> URL: http://tracker.firebirdsql.org/browse/CORE-4381
>> Project: Firebird Core
>>
> Sean, the solution is simple and elegant: Capture the transaction state
> information from the lead transaction using the transaction info
> mechanism then import it to secondary connection transactions with the
> transaction parameter block.
It requires two network transfers of (part
> 29.03.2014 20:06, Vlad Khorsun wrote:
>> 2. Correct place to fix: as you see above, we should fix VIO_verb_cleanup()
>> (not verb_post() as you points). Or we can change jrd_tra to allow it
>> to cache more
>> than one record buffer. Here i want to know opini
>> tra_undo_record was introduced especially to avoid frequent alloc\free
>> calls.
>> You offer to destroy it and revert old behavior. I don't like it.
>
> Take into account that the record content is copied from temporary storage
> (may be read
> from disk). Allocation from local pool
>> I'm against this patch, period.
>
> Nevertheless it solves the problem even if not in the best way. Because of
> that, I
> would suggest you to commit it now and rework later when you have some spare
> time.
The "problem" is far-far-far from critical and even from serious. You ev
> On 4/3/2014 11:58 AM, Vlad Khorsun wrote:
Jim.
I can't speak with you after your words. This is waste of our time.
Vlad
--
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.n
201 - 300 of 1273 matches
Mail list logo