Alex Peshkoff via Firebird-devel wrote 12.12.2021 18:52:
If it does not return sensitive information, I see no problem in add it to
examples UDR project.
With a check for SYSDBA I see no security risk with this UDR
Isn't GRANT EXECUTE to RDB$ADMIN enough? Or UDRs are not subject of SQL
r
Dmitry Yemanov wrote 08.12.2021 10:54:
For scrollable cursors, rows are prefetched and cached by the engine and thus
the count can be easily returned. It has its cost, but generally it's
unavoidable anyway for scrollable cursors.
Storing of fetched rows is unavoidable indeed but prefetch?..
Roman Simakov wrote 06.12.2021 18:50:
It looks so and we've returned to my original question)
Actually no. It is still unknown which version of Firebird 3.0 you watched
and which hash was used.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lis
Roman Simakov wrote 06.12.2021 13:16:
Can you point me to the commit? I see no difference between 3 and 5
versions of InternalHash::hash implementation.
It looks like using of CRC32C was backported into some 3.0.X. Original commit
was
https://github.com/FirebirdSQL/firebird/commit/a1d55c3e6
Dimitry Sibiryakov wrote 06.12.2021 12:26:
Hashing function has been changed to one with supposedly "more even"
distribution of values.
BTW a side effect of this change is that number of slots doesn't have to be
prime anymore.
--
WBR, SD.
Firebird-Devel mailing list, w
Roman Simakov wrote 06.12.2021 12:21:
3.0. Were there any changes in 4 or 5?
Yes. Hashing function has been changed to one with supposedly "more even"
distribution of values.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebir
Roman Simakov wrote 06.12.2021 10:24:
We can see:
Hash slots: 65521, Hash lengths (min/avg/max):0/ 13/ 30
Under heavy load.
On which version do you see it?
As long as the first number is zero increasing number of slots won't help.
--
WBR, SD.
Firebird-Devel mailing list, web inte
Mark Rotteveel wrote 01.12.2021 18:02:
You need to set the cursor name before you open it, IIRC.
Actually it doesn't matter. It is enough to set cursor name any time before
using of positional DML AFAIU.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourcefor
Dmitry Yemanov wrote 01.12.2021 16:06:
And I see the same problem for setCursorName() which is available only through
IStatement. An oversight?
IAttachment::openCursor() has a parameter for cursor name.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.
Dmitry Yemanov wrote 01.12.2021 9:10:
Given that any DSQL statement cannot have multiple result sets, I doubt
IResultSet::getInfo() is really required, IStatement::getInfo() could be used
instead. Or if we go for consistency, IResultSet::getInfo() could be added, but
internally it calls IStatem
Dmitry Yemanov wrote 30.11.2021 18:06:
Then reworking seems easy. Just preserve ISC_analyze_pclan() to extract the
lanman hostname and then try connecting via TCP.
The question is port number. It is changed more frequently than pipe name.
--
WBR, SD.
Firebird-Devel mailing list, web inte
Alex Peshkoff via Firebird-devel wrote 30.11.2021 17:57:
It can\t be reworked reliably - nobody can guarantee that DNS host name matches
lanman name.
Fortunately modern versions of gethostbyname() can resolve lanman names as
well.
is ability to mount network share on a letter still popular?
Leyne, Sean wrote 30.11.2021 17:15:
(We have 200+ servers deployed and none of them use (have never) WNET
With dropping this option ability to open database file from mounted network
drive also will be dropped (or at least must be heavily reworked).
--
WBR, SD.
Firebird-Devel mailing l
Dmitry Yemanov wrote 28.11.2021 12:51:
28.11.2021 14:47, Dimitry Sibiryakov wrote:
RDB$DB_KEY for the current record as well.
It's not a property of the cursor. Consider joins, unions, procedures, views,
etc.
It is a property of a current record the same as position. It is ok to r
Dmitry Yemanov wrote 28.11.2021 12:45:
That would be useful in some cases. And if statements with "WHERE CURRENT OF"
condition also were somehow marked client libraries could force cursor
position synchronization before executing them.
I'd expect FOR UPDATE to still disable batching even witho
Dmitry Yemanov wrote 28.11.2021 12:42:
Then we may support both "current position" and "total row count" in getInfo(),
RDB$DB_KEY for the current record as well.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Dmitry Yemanov wrote 28.11.2021 12:22:
Theoretically, we could extend IRecordSet with something similar (although it
would also require a protocol change), but the question is whether it's really
needed. Personally, I don't see it useful per se.
It could be useful for client-side caching dat
Dmitry Yemanov wrote 27.11.2021 12:23:
I would consider adding a NO_BATCH flag (which is currently triggered using FOR
UPDATE syntax) to cursorFlags.
That would be useful in some cases. And if statements with "WHERE CURRENT OF"
condition also were somehow marked client libraries could force
Tony Whyman wrote 26.11.2021 18:01:
Note that only if FGoToFirstRecordOnExecute is true will the returned values be
available immediately.
Which (accidentally) is default and has no sense to be changed ever.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourcef
Tony Whyman wrote 26.11.2021 17:13:
Legacy code can often appear "insane" with the benefit of hindsight. The problem
is that it exists and it is your starting point. You don't break it without good
reason.
Ok, here is legacy code from TIBSQL.ExecQuery:
case FSQLType of
SQLSelect:
Tony Whyman wrote 26.11.2021 16:32:
What you appear to be saying is that you have changed/expanded the semantic of
Update...Returning, changed the SQL Statement type returned and then not
expected the change to break any existing code...
Correction: any sane code.
If a singleton row is expe
Dimitry Sibiryakov wrote 26.11.2021 16:01:
4. With reference to item 1 above, if "Update..Returning" returns a statement
type of isc_info_sql_stmt_select then you _do_ have to also parse the
statement in IBX so that you know its an UPDATE and IStatement.execute has to
be called
Tony Whyman wrote 26.11.2021 15:28:
3. Prior to calling IStatement.Execute, the statement type is checked. The
current (IBX) code raises an exception if the sql statement type is
isc_info_sql_stmt_select in order to stop the wrong IStatement method being called.
So it is actually the test su
Tony Whyman wrote 26.11.2021 15:10:
This caused an error message to be generated because only a singleton row was
expected and not a cursor.
Well, all I can say is that "existing connectivity drivers do not support this
feature automagically". Is it really then intended behaviour for
Update..
Alex Peshkoff via Firebird-devel wrote 23.11.2021 18:09:
Yes, there is no special processing of them (so I can use them freely) and
my question is if it will be this way in the future.
No such plans but if you wan't better guarantees suggest to use double brackets,
i.e. [[ ]]
Double br
Alex Peshkoff via Firebird-devel wrote 23.11.2021 17:45:
square brackets? This one: [ ] ?
I do not remember and do not see in a code any special processing for them.
Yes, there is no special processing of them (so I can use them freely) and my
question is if it will be this way in the future
Hello All.
Currently Firebird config file parser accept a string in square brackets and
consider it to be a parameter without value.
May I rely on this behavior?
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Ivan Přenosil wrote 22.11.2021 12:21:
Thanks for you answers, but you are answering question I did not ask :)
In this case you should ask questions with a little more details. The answer
"everything is possible in soft world, it is just a matter of time and price"
isn't useful enough to be
Ivan Přenosil wrote 21.11.2021 14:22:
or could it be changed to read-only in FB4?
It is meaningless. Read-write transaction has no differences from read-only
until it perform an actual write.
And no, it cannot be "read-only read committed" for the consistency sake.
--
WBR, SD.
Firebi
Adriano dos Santos Fernandes wrote 09.11.2021 12:30:
Please note I'm refactoring the profiler branch, making it via a plugin.
Why? Do you expect anybody to write a custom profiler?
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/fir
Adriano dos Santos Fernandes wrote 08.11.2021 17:54:
The solution does not work for the profiler then.
Why? Isn't the profiler supposed to help detect cases when fast (by itself)
statement is executed enormous number of times degrading overall application
performance?..
--
WBR, SD.
Fi
Dmitry Yemanov wrote 08.11.2021 15:28:
Is it an important design of MON$STATEMENTS that this (subsequent
execution of a request maintain its ID) happens?
Nope, I believe the new ID must be generated also by findRequest() if the clone
was found in the cache.
Will it make impossible to det
Omacht András wrote 06.11.2021 14:08:
Maybe if you find double letters you handled them together when you sort and
you handled them separated when you search.
Unfortunately comparison is used for sorting and sorting is used for indexes
and indexes are used for search so they cannot have dif
Omacht András wrote 06.11.2021 12:35:
However, I am not sure whether it is not a better result for the average user
today, (e.g. in a quick search) if all appear after typing C...
AFAIU only "dictionary collations" are using contractions so developers of
applications for average users can u
Gabor Boros wrote 05.11.2021 16:39:
2021.11.05. 12:36 keltezéssel, Pavel Cisar írta:
However, the Czech language defines only one contraction (CH), but there are
other languages that have many such as Hungarian.
Yes: cs, dz, gy, ly, ny, sz, ty, zs and dzs. Sorry. :-)
Is it ok from the lan
Pro Turm wrote 01.11.2021 15:21:
My goal is indeed replication without a single point of failure. It the above
would have been possible (replication from two masters), than the following
would also have been possible :
H would be a replica in the case when nothing else works anymore, A,B,D have
Pro Turm wrote 01.11.2021 11:47:
Thanks. Could you please give some more words about this case?
Is the following constellation also possible?
A--\ /--E
B---D-F
C--/ \--G
That is that D replicates A,B,C (which share the same content e.g. A-B-C)
and simultaneo
Pro Turm wrote 01.11.2021 10:54:
Drop database C, recreate replica from scratch.
could you please say what the recreation from scratch would look like according
to your understanding ?
Quoting README.replication:
## Creating the replica database
In the Beta 1 release, any physica
Pro Turm wrote 27.10.2021 14:46:
And how about the following constellation:
The same. You still create a single failure point at B.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Pro Turm wrote 27.10.2021 14:39:
Which means cascading could be quite vulnerable. What is then the quickest
possible way to connect A to C in the above case?
Drop database C, recreate replica from scratch.
> Restarting the servers: are there some performance issues expected due to
>
Pro Turm wrote 27.10.2021 14:26:
How about the case A->B->C, and B fails? Will A be replicating directly to C ?
No.
What does a switch from primary <-> replica mean? Only changing the
configuration files?
No, it is invoking "gfix" command with appropriate switches as described in
RE
Pro Turm wrote 27.10.2021 13:52:
Is it possible and how does a replica become primary at some point?
As usual, changing of the replica mode of the database (if necessary at all).
Read readme.replication.txt in doc folder.
e.g. in A->B->C, when A fails to become B->C ?
It is already B-
Dmitry Yemanov wrote 27.10.2021 9:18:
Nope, it's configured on the replica side and allows the received changes to be
propagated further (if replica is also configured as a primary, i.e. A->B->C).
...and must be disabled for replication A->B->A i.e. bidirectional to prevent
infinite data bou
Mark Rotteveel wrote 23.10.2021 14:31:
I'm not sure why not.
Because of record buffering it is too hard for a feature with so little usage.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Tony Whyman wrote 23.10.2021 14:04:
Is this true?
Yes. This feature is not implemented in standard network plugin and so far
nobody created a custom one.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Vlad Khorsun wrote 14.10.2021 18:07:
# To enter curvy brackets { } somewhere in a configuration dup them: {{ }}.
Hope it helps
Yes, it did. Thanks.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Hello All,
Is there a way to workaround "illegal line" error if curly brackets are part
of database GUID used as a key name:
{F5B90BDE-6DA3-4095-A9E1-6D26AE34519F} = target.fdb
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/fire
Adriano dos Santos Fernandes wrote 14.10.2021 15:23:
This happens because transaction 3 will fill the metadata caches.
Can we consider this trick safe?
Even if it is safe, it is a violation of "snapshot TIL" contract IMHO and
must be fixed.
I do want to use it in this situation:
Profile
Vlad Khorsun wrote 13.10.2021 14:39:
Due to this, I see no reason to replicate physical
layout of master database to the replica. I.e. TABLESPACE-related statements
should not be replicated at all, IMO. Other opinions ?
I fully agree with not replicated "CREATE/ALTER/DROP TABLESPACE" but what
Roman Simakov wrote 13.10.2021 14:20:
I suppose you mean async replication. I guess like other errors, no?
No, I meant exactly synch replication. "Other errors" in such cases leads to
transaction rollback on both sides. Can tablespace moving to be rolled back?
--
WBR, SD.
Firebird-Deve
About moving of tablespaces and objects between tablespaces: on primary site
it succeeded, on synchronous replica it failed (and vice versa). How such
situation is going to be handled?
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/
Roman Simakov wrote 12.10.2021 16:30:
It allows you to restore tablespace contents to new places.
If the option is not specified gbak will use old locations for every
tablespace.
I still think this should be an explicit option as well, otherwise
surprising things - for the user - can happen when
Roman Simakov wrote 12.10.2021 17:51:> More details you will find in PR and then
we will be able to discuss it.
Actually we don't discuss. I just beg you for explanation why it was done
this way, which alternatives were considered and why they were found
unacceptable or worse.
Such explana
Dmitry Yemanov wrote 12.10.2021 17:34:
They're modified in user transaction(s).
Nevertheless users have no right for direct modifications so they cannot mess
with the page number value. If we exclude the field from the query inside of
AlterXXX methods it should to be safe, right?
--
WBR
Roman Simakov wrote 12.10.2021 17:16:
I'm not sure that keeping page numbers in a transactional relation is
a really excellent idea.
But you added it into RDB$RELATIONS table, no?..
System tables are operated in system transaction which effectively uses TIL
dirty read so there is no real v
Dmitry Yemanov wrote 12.10.2021 16:59:
Slightly faster inserts into indices.
If local indexes are used, yes. But such indexes require partitioning key in
queries.
What will be effect for global indexes? Reference to the partition for a
record has to be in index node, no?
But it may be
Roman Simakov wrote 12.10.2021 16:55:
If this field is needed at all. Still there is no answer why tablespace
cannot be identified by name only except "it was done this way 40 years ago".
Why not to suggest PR for this refactoring?
I understand that you insist on merging your existing co
Dmitry Yemanov wrote 12.10.2021 15:02:
Partition is a page set. Different page sets may be surely stored inside a
single database file, but they may also be stored in different files. It makes a
lot of sense to nbackup only some partition(s), or store them on slow drives,
etc -- the same as for
Dmitry Yemanov wrote 12.10.2021 14:19:
There's also 4th goal: provide internal infrastructure (splitting page spaces
into multiple files) to support partitioning later. I expect these two features
to share a lot.
In this case your vision of partitioning is quite special because in others'
i
Roman Simakov wrote 11.10.2021 14:17:
GOALS
==
1) Extend the current limits on database size
Current limit is 16 TB and can be extended without explicit tablespace
managing by something similar to OS memory mapping technique effectively adding
some external-sourced bits to current 32 bi
Vlad Khorsun wrote 12.10.2021 12:10:
I see no problem with increasing the limit. I see problems with reducing it
(someone may use them). So let's start from a small number 63. When we
implement partitions we increase it more consciously.
I speak about data type used in ODS for tablespace ID
Mark Rotteveel wrote 12.10.2021 10:55:
It allows you to restore tablespace contents to new places.
If the option is not specified gbak will use old locations for every tablespace.
I still think this should be an explicit option as well, otherwise surprising
things - for the user - can happen w
Roman Simakov wrote 11.10.2021 20:23:
> pag_header in every tablespace is reserved and may be replaced by a
> new page type.
You mean page zero, which is currently always pag_header. I see no
reason
to change this, so far. Header page uses to describe properties of databas
Roman Simakov wrote 11.10.2021 20:23:
> Note: *MAIN* - is a name of the basic database file.
Please, use *DEFAULT* for default (main) tablespace at "main" database
file.
It is much more consistent with SQL and allows to avoid new unnecessary
keyword.
I'd be happy to agree. A
Dmitry Yemanov wrote 11.10.2021 18:24:
It's possible to create up to 253 tablespaces.
Two ID's is reserved for DEFAULT and TEMPORARY tablespaces, correct ?
I would reserve some more ID's for future system usage. I don't see it as
limitation for end users.
For regular tablespaces (created e
Hello All.
In which case charset received by IReplTransaction::executeSqlIntl() can be
different from charset of attachment received by IReplicatedSession::init()?
I can imagine only the case of cascade replication when the Applier hacks
replicator's connection charset.
--
WBR, SD.
Roman Simakov wrote 11.10.2021 15:23:
Cannot be tablespace identification loaded from RDB$INDEXES?
Adding it into irp will increase size of irt_repeat by 1/3 decreasing limit
of indexes per table, no?
I'm sure the answer is the same as why the root page cannot be loaded
from RDB$INDEXES
Roman Simakov wrote 11.10.2021 14:52:
Add page space id to page number in ods.h:index_root_page.
pag_root is located in the tablespace where a table is located
Why is this difference? What does prevent you from putting irp into the
tablespace where index is located?
IRP describes every inde
Dimitry Sibiryakov wrote 11.10.2021 14:26:
What's the point of using "MAIN" here?
Nevemind, I've found the answer.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Roman Simakov wrote 11.10.2021 14:17:
FIELD TYPE CONSTRAINT ... USING INDEX ... TABLESPACE { | MAIN} --
field constraint tablespace
What's the point of using "MAIN" here? Whole TABLESPACE clause cannot be
omitted?
Add page space id to page number in ods.h:index_root_page.
pag_root is
Mark Rotteveel wrote 10.10.2021 18:08:
Yes, this is normal. The range of a smallint is from -32768 to 32767, so you
can't negate -32768 as it will overflow back to -32768, hence it will raise the
integer overflow error.
The question is whether result of negation should keep the type of sourc
Mark Rotteveel wrote 10.10.2021 16:03:
Could you please stop posting random links without further context. And as far
as I can tell that blogger is just a random person which has no authority to
'declare' SRP deprecated.
Besides, Firebird uses SRP for authentication, not just key exchange.
Hello All.
Currently Firebird is unable to load plugins via symlinks. Is it intentional?
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Mark Rotteveel wrote 07.10.2021 16:11:
Backing up only one tablespace could lead to inconsistent restores though.
From practical POV it is not different from current partial backup including
only subset of tables.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists
Roman Simakov wrote 07.10.2021 16:08:
Let's eat an elephant one bite at a time
Yes, that's exactly my point: no need to reserve fields if implementation
that use them is not clear yet and better no feature than provisory hack.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
Molnár Attila wrote 07.10.2021 10:56:
It might not be about read performance, but e.g. logical data serparation,
backup speedup, etc...
Logical data separation is task for namespaces, not tablespaces.
Physical backup of single tablespace is pointless because it has no TIP so
without main
Roman Simakov wrote 06.10.2021 21:49:
How are data moved?
Under transaction control?
In background (and interruptible) like db crypt?
it's a dfw operation with EX database lock. I hope to relax this
limitation later. So there are no concurrent changes.
1) copy all data pages
2) switch RDB$POIN
Roman Simakov wrote 07.10.2021 8:14:
ср, 6 окт. 2021 г. в 23:57, Dimitry Sibiryakov:
Roman Simakov wrote 06.10.2021 22:44:
Sorry I was wrong. TABLESPACE_ID is used as pagespace ID internally
but this id presents at index root page for example.
I don't see it in ods.h:index_root
Roman Simakov wrote 06.10.2021 22:44:
Sorry I was wrong. TABLESPACE_ID is used as pagespace ID internally
but this id presents at index root page for example.
I don't see it in ods.h:index_root_page.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net
Roman Simakov wrote 06.10.2021 21:27:
So am I. Currently tablespace headers are just not used.
In this case how do you determine page size inside of tablespace file? Having
it different from main one is an essential feature for indexes.
--
WBR, SD.
Firebird-Devel mailing list, web inte
Roman Simakov wrote 06.10.2021 20:43:
Let me not describe every field. I hope most of them are obvious.
No, they aren't.
ID is not referenced anywhere and seems to have no purpose at all.
TABLESPACE_NAME being CHAR is pointless. CHAR was used in ancient times but
new fields should prefer
Roman Simakov wrote 06.10.2021 17:32:
A new table RDB$TABLESPACES:
RDB$TABLESPACE_ID - SMALLINT
RDB$TABLESPACE_NAME - CHAR (63)
RDB$SECURITY_CLASS - CHAR (63)
RDB$SYSTEM_FLAG - SMALLINT
RDB$DESCRIPTION - BLOB SUBTYPE TEXT SEGMENT SIZE 80
RDB$OWNER_NAME - CHAR (63)
RDB$FILE_NAME - V
Nils Bödeker wrote 04.10.2021 13:30:
Possible Cause
-
I inserting an Record in the table BINARYDATA an BEFORE_INSERT Trigger Insert
an additional Record in the table OBJECTS. If i disable this Trigger the Error
don't happen.
DML triggers don't fire during replication.
Fir
Alex Peshkoff via Firebird-devel wrote 21.09.2021 15:06:
Valgring makes execution many times slower (that's from my own experience).
Does it really matter if test suite isn't run in production anyway?..
Can you check that for performance?
No. They are not available for Windows as well a
Alex Peshkoff via Firebird-devel wrote 21.09.2021 14:51:
That's why this solution is usable only if we can't do something else.
https://github.com/google/sanitizers can detect much wider set of problems
that our pools. Its support is built into GCC.
--
WBR, SD.
Firebird-Devel mailing l
Adriano dos Santos Fernandes wrote 21.09.2021 14:24:
Though we will lose way to monitor leaks.
Valgrind, drMemory and sanitizers will work in this case.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
Adriano dos Santos Fernandes wrote 21.09.2021 13:50:
We may split that global default pool in two, so we'll have another
default pool for external (not controlled by us) libraries. All memory
allocated directly with "operator new" (not FB_NEW) will go to external
pool.
Wouldn't it be simpler
Alex Peshkoff via Firebird-devel wrote 17.09.2021 17:03:
as an additional cherry on a cake we have very fast, efficient and (last but not
least) familiar for developers support of hunting for memory leaks.
Is it more familiar than DrMemory and valgrind? Former is currently useless
because of
Alex Peshkoff via Firebird-devel wrote 17.09.2021 16:13:
First of all dynamically loaded libraries are unloaded in reverse order, next
invoken executable dtors/atexits, and only after it - dtors of linked libraries.
Ah, ok. But returning to the topic: which library must provide pooled memory
Alex Peshkoff via Firebird-devel wrote 17.09.2021 9:08:
Is it about a library dynamically (dlopen, LoadLibrary) loaded or linked to the
exacutable?
Both. The library tree is the same (at least on Linux).
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sourceforge
Adriano dos Santos Fernandes wrote 16.09.2021 21:43:
And it should happen if you called it before de constructors. Or it was
not done in this way?
I didn't call it at all. Firebird engine did it for me.
Are there any published test? Did it was compiled with c++11?
Look at whole thread:
Adriano dos Santos Fernandes wrote 16.09.2021 20:39:
I wanted to control the order of constructors mostly to control the order of
destructors (which should be inverse).
KISS. Just gather them all into single CPP file.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lis
Adriano dos Santos Fernandes wrote 16.09.2021 21:08:
And this has 0 importance for the matter, considering that tools vendors
implement what the standard describes.
We already had a discussion here that shown that atexit (and called from it
PluginModule::doCleanup()) is fired after global de
Adriano dos Santos Fernandes wrote 16.09.2021 20:39:
It looks like there is another way to make that.
std::atexit (https://en.cppreference.com/w/cpp/utility/program/atexit) docs
says:
No matter what docs say, std::atexit is just a synonym for atexit.
--
WBR, SD.
Firebird-Devel mailing
Alex Peshkoff via Firebird-devel wrote 16.09.2021 12:21:
In new version we can easily change MessageBuffer format.
Or we can return XSQLDA to OO API and deprecate MessageBuffer repeating
evolution of ISC API.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
https://lists.sour
Alex Peshkoff via Firebird-devel wrote 16.09.2021 12:21:
In new version we can easily change MessageBuffer format. The main issue is old
user's binaries - probably to support them we will need some tricks in remote
redirector & copy/in/out in dsql.cpp.
This trick could be "use offsets from p
Alex Peshkoff via Firebird-devel wrote 16.09.2021 11:52:
First of all it's good to know what places are affected. Do not want to try to
guess them.
ODS and MessageBuffer at least. Record is an internal thing and can be
ignored.
--
WBR, SD.
Firebird-Devel mailing list, web interface at
Alex Peshkoff via Firebird-devel wrote 13.09.2021 9:39:
Yes, that should be fixed. If one builds FB with native __int128 it fails on
unaligned data access. But changing alignment requirement to 16 bytes is not
enough - appears some buffers are not aligned at 16-bit cause FB fails exactly
in t
Maya Opperman wrote 13.09.2021 15:59:
Does this need to be done within every transaction that is started, or per
connection/session, or once off per database?
Quoting the readme:
Implements capability to setup columns coercion rules *in current session*.
--
WBR, SD.
Firebird-Devel
Alex Peshkoff via Firebird-devel wrote 15.05.2018 12:13:
On 05/15/18 13:06, Dimitry Sibiryakov wrote:
15.05.2018 11:58, Alex Peshkoff via Firebird-devel wrote:
Do alignment of fields in C struct matches alignment of the same fields in
MessageBuffer?
If at compile/preprocess time construct
101 - 200 of 2959 matches
Mail list logo
