Re: Bootstrap download failures, mainly from http://sourceforge.net/projects/oooextras.mirror/files

2015-06-23 Thread Roberto Galoppini
2015-06-23 0:49 GMT+02:00 Kay Schenk kay.sch...@gmail.com:



 On 06/21/2015 11:54 PM, Andrea Pescetti wrote:
  On 21/06/2015 Kay Schenk wrote:
  yes, they would be...based on the buildbot failure alone, I think the SF
  mirror must have been deactivated just a day or so ago.
 
  Could it be that this is an unintended consequence of SourceForge's new
  policy when they decided, for transparency towards the community, to
  avoid mirrors that had not been asked for?
 
 http://sourceforge.net/blog/project-mirroring-policies-will-be-revisited-with-our-community-panel-existing-mirrors-removed/
 
 
  It seems clear that
  http://sourceforge.net/projects/oooextras.mirror/files/ was definitely
  to be kept (or, while at it, we could have it relocated to
  http://sourceforge.net/projects/oooextras/ or something else like
  apache-extras-openoffice).

 Oh my! I did not see/hear about this. Given the June 18 notice, yes, I
 would certainly say this applies to AOO and why the mirror service was
 stopped.


Yes, that is the reason. Do we need to reinstate the project, and if that's
the case who would be the administrator/s?
Let me know and I'll get it fixed, really sorry for the annoyance.

Roberto




 OK, I'm looking specifically at this paragraph from the link you sent--

 Mirrors which are not co-maintained with the one or more of the
 original developers, except where the upstream site has been
 discontinued, have been removed effective immediately. An extensive
 review has been conducted of the 295 mirrored projects and removals were
 completed on June 18th. Where a SourceForge-maintained mirror has been
 removed, we have redirected this traffic back to the canonical home for
 the software (whence it was mirrored to begin with).

 Basically since ooo-extras are copies with md5 sums applied for our
 own uses, and we are NOT the original developers,  this would definitely
 apply.

 In most cases, we have the provider links to these items listed as URL1
 in /main/external_deps.lst. So, should we just change this file to use
 URL1 as default and forget ooo-extras altogether?

 We would need to see how this affects configuration which I've had
 problems with in the past when these sources haven't bee kept up to date
 in /ext_sources. It's likely we would need some alternate verification
 method.




 
  The direct download from
  http://www.ijg.org/files/jpegsrc.v8d.tar.gz
  and
  http://www.python.org/ftp/python/2.7.6/Python-2.7.6.tgz
  works for me. I don't know why it doesn't work for bootstrap.
 
  At times, the first download fails since the .gz file is uncompressed
  during download. So basically you download file.tar.gz and what you get
  is a file named file.tar.gz but with the contents of file.tar (and
  thus a different md5sum). If someone wants to hack the Perl file or file
  this in an issue it would be perfect. But the fallback always worked, so
  this hasn't been high priority.
 
  Regards,
Andrea.
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
  For additional commands, e-mail: dev-h...@openoffice.apache.org
 

 --
 
 MzK

 We can all sleep easy at night knowing that
  somewhere at any given time,
  the Foo Fighters are out there fighting Foo.
   -- David Letterman

 -
 To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
 For additional commands, e-mail: dev-h...@openoffice.apache.org




review requested: [Issue 105098] [Mac] inserting files (e. g. images) into a document by drag drop isn't possible : [Attachment 84803] Patch for drag n drop bug with rev #2

2015-06-23 Thread bugzilla
h...@apache.org h...@apache.org has reassigned Manik mm...@yahoo.com's
request for review to h...@apache.org h...@apache.org:
Issue 105098: [Mac] inserting files (e. g. images) into a document by drag 
drop isn't possible
https://bz.apache.org/ooo/show_bug.cgi?id=105098

Attachment 84803: Patch for drag n drop bug with rev #2
https://bz.apache.org/ooo/attachment.cgi?id=84803action=edit



--- Comment #23 from h...@apache.org h...@apache.org ---
Comment on attachment 84803
  -- https://bz.apache.org/ooo/attachment.cgi?id=84803
Patch for drag n drop bug with rev #2

reviewing the patch...

-
To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
For additional commands, e-mail: dev-h...@openoffice.apache.org



Re: Looking for help for clipboard in Math

2015-06-23 Thread Regina Henschel

Hi Michael,

and another report:

I found, what I did wrong. I had used GetTree() and not realized, that 
it does not only return a pointer but actually performs a pop.
For text clipboard content I have followed your tip with SvMemoryStream 
and utl::OInputStreamWrapper.
So now I can import from clipboard, if the clipboard is of MIME type 
application/mathml+xml or if it is of type FORMAT_STRING and is 
actually a math-fragment.


But now I have this problem:
I start module Math with a fresh document, I write a formula into the 
command window (e.g. a^2+b^2) and then try to save as *.mml.
I get the errors cited below. I have not used any operations with the 
clipboard. A build from current trunk does not have this error, so it 
must be something, which I have introduced. But I do not know, what are 
possible causes and where to start.


Errormessages:

Locking problem.
Sharing violation while accessing the object.


Error: An error or a warning has occurred during XML import/export!
Error-Id: 0x60040004
Flags: 6 ERRROR SEVERE
Class: 4 API
Number: 4
Parameters:
Exception-Message: io exception during writing
 From File 
c:/AOO_debugbuild_420/trunk/main/xmloff/source/core/xmlerror.cxx at Line 193


 Error: cached Sequence not written From File 
C:/AOO_debugbuild_420/trunk/main/sax/source/expatwrap/saxwriter.cxx at 
Line 164



Kind regards
Regina


Michael Stahl schrieb:

On 06.06.2015 22:03, Regina Henschel wrote:

Hi Michael,

thank-you for looking at my problems.

Michael Stahl schrieb:

On 05.06.2015 01:44, Regina Henschel wrote:

Hi all,

I struggle with the clipboard. My goal is to import MathML in module
Math from clipboard, similar as it is imported from file. But I'm stuck.
Therefore some questions:

[skipped a lot of text]

Helpful comments. Looking around I think, the connection to the OS is
done in /main/dtrans/source/win32/

I have found the table m_TranslTable, which refers the SOT_FORMATSTR_IDs
from exchange.cxx. I'll try what happens, when I add the MathML format
there too. I had already added it in exchange.cxx, but that was not
enough to be recognized. [I need some time for that. I will report back,
when I have finished.]


that sounds plausible.


(5)
I try to use SmViewShell::InsertFrom(SfxMedium rMedium). It seems to
work, but when the process arrives at SmXMLImport::endDocument(void),
the node tree is empty.  Any tips, what I might have missed?


how do you prepare the SfxMedium?  i'm not sure if it requires an actual
file, or if it can read from an input stream (XInputStream); it probably
can't read from a Sequencesal_Int8 buffer directly...

there is a SfxMedium::setStreamToLoadFrom() which looks promising.


I have used:

TransferableDataHelper aDataHelper(
TransferableDataHelper::CreateFromSystemClipboard(GetEditWindow()) );
uno::Reference  io::XInputStream  xStrm;
aDataHelper.GetInputStream( nId, xStrm );
SfxMedium* pClipboardMedium = new SfxMedium();
SfxMedium aClipboardMedium = *pClipboardMedium;


this is probably not a good idea: the aClipboardMedium is now a copy of
pClipboardMedium (using the copy-constructor), so any modification to
aClipboardMedium will not be visible if you use pClipboardMedium.

if you need a pointer, better do it the other way around, and use 
aClipbardMedium when needed.


aClipboardMedium.setStreamToLoadFrom( xStrm, sal_True /*bIsReadOnly*/ );
InsertFrom(aClipboardMedium);


oh, nice, the TransferableDataHelper can already get you a stream.


When I then proof it with

SvStream* pStream = aClipboardMedium.GetInStream();
...
sal_uLong nBytesRead = pStream-Read( aBuffer, nBufferSize );
printf(%s \n, aBuffer);

I can see, that the stream contains the expected MathML-source in case
the clipboard viewer lists the clipboard format application/mathml+xml.


ok, so we can read the data.

do you re-wind the stream with Seek() after this debug output?  perhaps
the import filter reads from the current position, which is going to be
in the middle or at the end of the stream after this.


In addition I have set the filter by

const SfxFilter* pMathFilter = SfxFilter::GetFilterByName(
String::CreateFromAscii(MATHML_XML) );
aClipboardMedium.SetFilter(pMathFilter);

so that

if ( rFltName.EqualsAscii(MATHML_XML) )

in InsertFrom becomes true


it looks like SmXMLImportWrapper is using the usual xmloff XML-parsing
stuff.

so i would try to check if the root element of the MathML document is
being recognized; set a breakpoint or add a SAL_DEBUG output in the
right CreateChildContext() method...

this is apparently SmXMLImport::CreateContext() - which should create a
SmXMLDocContext_Impl, since there's not going to be a office namespace.

most of the work is going on in the CreateChildContext() and
EndElement() overrides in various child classes of SmXMLImportContext;
basically xmloff maintains a stack of contexts, one for each currently
open XML element; it then calls StartElement() / CreateChildContext() /
EndElement() etc. on the context that is on top of 

Re: Bootstrap download failures, mainly from http://sourceforge.net/projects/oooextras.mirror/files

2015-06-23 Thread Kay Schenk

On 06/23/2015 08:00 AM, Roberto Galoppini wrote:
 2015-06-23 0:49 GMT+02:00 Kay Schenk kay.sch...@gmail.com:
 


 On 06/21/2015 11:54 PM, Andrea Pescetti wrote:
 On 21/06/2015 Kay Schenk wrote:
 yes, they would be...based on the buildbot failure alone, I think the SF
 mirror must have been deactivated just a day or so ago.

 Could it be that this is an unintended consequence of SourceForge's new
 policy when they decided, for transparency towards the community, to
 avoid mirrors that had not been asked for?

 http://sourceforge.net/blog/project-mirroring-policies-will-be-revisited-with-our-community-panel-existing-mirrors-removed/


 It seems clear that
 http://sourceforge.net/projects/oooextras.mirror/files/ was definitely
 to be kept (or, while at it, we could have it relocated to
 http://sourceforge.net/projects/oooextras/ or something else like
 apache-extras-openoffice).

 Oh my! I did not see/hear about this. Given the June 18 notice, yes, I
 would certainly say this applies to AOO and why the mirror service was
 stopped.

 
 Yes, that is the reason. Do we need to reinstate the project, and if that's
 the case who would be the administrator/s?
 Let me know and I'll get it fixed, really sorry for the annoyance.
 
 Roberto

In the short term, reinstating the mirror of /trunk/ext_sources is
pretty important without a lot of additional testing/setup directly from
the source repositories of these items. So, I would vote YES on this.

What is required of the administrator(s)?

No apology needed. I think the majority of us are not current on
SourceForge policies so we were not aware of this change. As far as I
know, Apache OpenOffice was happy to have the SourceForge mirror service
for these additional sources.

 
 
 

 OK, I'm looking specifically at this paragraph from the link you sent--

 Mirrors which are not co-maintained with the one or more of the
 original developers, except where the upstream site has been
 discontinued, have been removed effective immediately. An extensive
 review has been conducted of the 295 mirrored projects and removals were
 completed on June 18th. Where a SourceForge-maintained mirror has been
 removed, we have redirected this traffic back to the canonical home for
 the software (whence it was mirrored to begin with).

 Basically since ooo-extras are copies with md5 sums applied for our
 own uses, and we are NOT the original developers,  this would definitely
 apply.

 In most cases, we have the provider links to these items listed as URL1
 in /main/external_deps.lst. So, should we just change this file to use
 URL1 as default and forget ooo-extras altogether?

 We would need to see how this affects configuration which I've had
 problems with in the past when these sources haven't bee kept up to date
 in /ext_sources. It's likely we would need some alternate verification
 method.





 The direct download from
 http://www.ijg.org/files/jpegsrc.v8d.tar.gz
 and
 http://www.python.org/ftp/python/2.7.6/Python-2.7.6.tgz
 works for me. I don't know why it doesn't work for bootstrap.

 At times, the first download fails since the .gz file is uncompressed
 during download. So basically you download file.tar.gz and what you get
 is a file named file.tar.gz but with the contents of file.tar (and
 thus a different md5sum). If someone wants to hack the Perl file or file
 this in an issue it would be perfect. But the fallback always worked, so
 this hasn't been high priority.

 Regards,
   Andrea.

 -
 To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
 For additional commands, e-mail: dev-h...@openoffice.apache.org


 --
 
 MzK

 We can all sleep easy at night knowing that
  somewhere at any given time,
  the Foo Fighters are out there fighting Foo.
   -- David Letterman

 -
 To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
 For additional commands, e-mail: dev-h...@openoffice.apache.org


 

-- 

MzK

We can all sleep easy at night knowing that
 somewhere at any given time,
 the Foo Fighters are out there fighting Foo.
  -- David Letterman

-
To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
For additional commands, e-mail: dev-h...@openoffice.apache.org



Re: Bootstrap download failures, mainly from http://sourceforge.net/projects/oooextras.mirror/files

2015-06-23 Thread Roberto Galoppini
2015-06-23 22:39 GMT+02:00 Kay Schenk kay.sch...@gmail.com:


 On 06/23/2015 08:00 AM, Roberto Galoppini wrote:
  2015-06-23 0:49 GMT+02:00 Kay Schenk kay.sch...@gmail.com:
 
 
 
  On 06/21/2015 11:54 PM, Andrea Pescetti wrote:
  On 21/06/2015 Kay Schenk wrote:
  yes, they would be...based on the buildbot failure alone, I think the
 SF
  mirror must have been deactivated just a day or so ago.
 
  Could it be that this is an unintended consequence of SourceForge's new
  policy when they decided, for transparency towards the community, to
  avoid mirrors that had not been asked for?
 
 
 http://sourceforge.net/blog/project-mirroring-policies-will-be-revisited-with-our-community-panel-existing-mirrors-removed/
 
 
  It seems clear that
  http://sourceforge.net/projects/oooextras.mirror/files/ was definitely
  to be kept (or, while at it, we could have it relocated to
  http://sourceforge.net/projects/oooextras/ or something else like
  apache-extras-openoffice).
 
  Oh my! I did not see/hear about this. Given the June 18 notice, yes, I
  would certainly say this applies to AOO and why the mirror service was
  stopped.
 
 
  Yes, that is the reason. Do we need to reinstate the project, and if
 that's
  the case who would be the administrator/s?
  Let me know and I'll get it fixed, really sorry for the annoyance.
 
  Roberto

 In the short term, reinstating the mirror of /trunk/ext_sources is
 pretty important without a lot of additional testing/setup directly from
 the source repositories of these items. So, I would vote YES on this.



oooextras is back from few hours now.



 What is required of the administrator(s)?



A sourceforge account, let me know yours (even privately) and I'll add them
right away.



 No apology needed. I think the majority of us are not current on
 SourceForge policies so we were not aware of this change. As far as I
 know, Apache OpenOffice was happy to have the SourceForge mirror service
 for these additional sources.


Our decision was driven by other communities' concerns, we believe that our
current setting give peace of mind to all mirrored projects, but those who
really want to manage their mirrors! :) Hope we reinstated it properly and
we can move on.

Thanks for understanding.

roberto





 
 
 
 
  OK, I'm looking specifically at this paragraph from the link you sent--
 
  Mirrors which are not co-maintained with the one or more of the
  original developers, except where the upstream site has been
  discontinued, have been removed effective immediately. An extensive
  review has been conducted of the 295 mirrored projects and removals were
  completed on June 18th. Where a SourceForge-maintained mirror has been
  removed, we have redirected this traffic back to the canonical home for
  the software (whence it was mirrored to begin with).
 
  Basically since ooo-extras are copies with md5 sums applied for our
  own uses, and we are NOT the original developers,  this would definitely
  apply.
 
  In most cases, we have the provider links to these items listed as URL1
  in /main/external_deps.lst. So, should we just change this file to use
  URL1 as default and forget ooo-extras altogether?
 
  We would need to see how this affects configuration which I've had
  problems with in the past when these sources haven't bee kept up to date
  in /ext_sources. It's likely we would need some alternate verification
  method.
 
 
 
 
 
  The direct download from
  http://www.ijg.org/files/jpegsrc.v8d.tar.gz
  and
  http://www.python.org/ftp/python/2.7.6/Python-2.7.6.tgz
  works for me. I don't know why it doesn't work for bootstrap.
 
  At times, the first download fails since the .gz file is uncompressed
  during download. So basically you download file.tar.gz and what you get
  is a file named file.tar.gz but with the contents of file.tar (and
  thus a different md5sum). If someone wants to hack the Perl file or
 file
  this in an issue it would be perfect. But the fallback always worked,
 so
  this hasn't been high priority.
 
  Regards,
Andrea.
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
  For additional commands, e-mail: dev-h...@openoffice.apache.org
 
 
  --
  
  MzK
 
  We can all sleep easy at night knowing that
   somewhere at any given time,
   the Foo Fighters are out there fighting Foo.
-- David Letterman
 
  -
  To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
  For additional commands, e-mail: dev-h...@openoffice.apache.org
 
 
 

 --
 
 MzK

 We can all sleep easy at night knowing that
  somewhere at any given time,
  the Foo Fighters are out there fighting Foo.
   -- David Letterman

 -
 

Re: Bootstrap download failures, mainly from http://sourceforge.net/projects/oooextras.mirror/files

2015-06-23 Thread Kay Schenk

On 06/23/2015 01:48 PM, Roberto Galoppini wrote:
 2015-06-23 22:39 GMT+02:00 Kay Schenk kay.sch...@gmail.com:
 

 On 06/23/2015 08:00 AM, Roberto Galoppini wrote:
 2015-06-23 0:49 GMT+02:00 Kay Schenk kay.sch...@gmail.com:



 On 06/21/2015 11:54 PM, Andrea Pescetti wrote:
 On 21/06/2015 Kay Schenk wrote:
 yes, they would be...based on the buildbot failure alone, I think the
 SF
 mirror must have been deactivated just a day or so ago.

 Could it be that this is an unintended consequence of SourceForge's new
 policy when they decided, for transparency towards the community, to
 avoid mirrors that had not been asked for?


 http://sourceforge.net/blog/project-mirroring-policies-will-be-revisited-with-our-community-panel-existing-mirrors-removed/


 It seems clear that
 http://sourceforge.net/projects/oooextras.mirror/files/ was definitely
 to be kept (or, while at it, we could have it relocated to
 http://sourceforge.net/projects/oooextras/ or something else like
 apache-extras-openoffice).

 Oh my! I did not see/hear about this. Given the June 18 notice, yes, I
 would certainly say this applies to AOO and why the mirror service was
 stopped.


 Yes, that is the reason. Do we need to reinstate the project, and if
 that's
 the case who would be the administrator/s?
 Let me know and I'll get it fixed, really sorry for the annoyance.

 Roberto

 In the short term, reinstating the mirror of /trunk/ext_sources is
 pretty important without a lot of additional testing/setup directly from
 the source repositories of these items. So, I would vote YES on this.

 
 
 oooextras is back from few hours now.

Ok, thanks! :)

 
 

 What is required of the administrator(s)?

 
 
 A sourceforge account, let me know yours (even privately) and I'll add them
 right away.

I will contact you privately about this.

 
 

 No apology needed. I think the majority of us are not current on
 SourceForge policies so we were not aware of this change. As far as I
 know, Apache OpenOffice was happy to have the SourceForge mirror service
 for these additional sources.

 
 Our decision was driven by other communities' concerns, we believe that our
 current setting give peace of mind to all mirrored projects, but those who
 really want to manage their mirrors! :) Hope we reinstated it properly and
 we can move on.
 
 Thanks for understanding.
 
 roberto

OK, thank you.

 
 
 
 





 OK, I'm looking specifically at this paragraph from the link you sent--

 Mirrors which are not co-maintained with the one or more of the
 original developers, except where the upstream site has been
 discontinued, have been removed effective immediately. An extensive
 review has been conducted of the 295 mirrored projects and removals were
 completed on June 18th. Where a SourceForge-maintained mirror has been
 removed, we have redirected this traffic back to the canonical home for
 the software (whence it was mirrored to begin with).

 Basically since ooo-extras are copies with md5 sums applied for our
 own uses, and we are NOT the original developers,  this would definitely
 apply.

 In most cases, we have the provider links to these items listed as URL1
 in /main/external_deps.lst. So, should we just change this file to use
 URL1 as default and forget ooo-extras altogether?

 We would need to see how this affects configuration which I've had
 problems with in the past when these sources haven't bee kept up to date
 in /ext_sources. It's likely we would need some alternate verification
 method.





 The direct download from
 http://www.ijg.org/files/jpegsrc.v8d.tar.gz
 and
 http://www.python.org/ftp/python/2.7.6/Python-2.7.6.tgz
 works for me. I don't know why it doesn't work for bootstrap.

 At times, the first download fails since the .gz file is uncompressed
 during download. So basically you download file.tar.gz and what you get
 is a file named file.tar.gz but with the contents of file.tar (and
 thus a different md5sum). If someone wants to hack the Perl file or
 file
 this in an issue it would be perfect. But the fallback always worked,
 so
 this hasn't been high priority.

 Regards,
   Andrea.

 -
 To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
 For additional commands, e-mail: dev-h...@openoffice.apache.org


 --
 
 MzK

 We can all sleep easy at night knowing that
  somewhere at any given time,
  the Foo Fighters are out there fighting Foo.
   -- David Letterman

 -
 To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
 For additional commands, e-mail: dev-h...@openoffice.apache.org




 --
 
 MzK

 We can all sleep easy at night knowing that
  somewhere at any given time,
  the Foo Fighters are out there fighting Foo.
   -- David Letterman

 

Re: Looking for help for clipboard in Math

2015-06-23 Thread Michael Stahl
On 23.06.2015 20:46, Regina Henschel wrote:
 But now I have this problem:
 I start module Math with a fresh document, I write a formula into the 
 command window (e.g. a^2+b^2) and then try to save as *.mml.
 I get the errors cited below. I have not used any operations with the 
 clipboard. A build from current trunk does not have this error, so it 
 must be something, which I have introduced. But I do not know, what are 
 possible causes and where to start.
 
 Errormessages:
 
 Locking problem.
 Sharing violation while accessing the object.

this is a problem with file locking that generally only manifests on
Windows: a file has been opened once and is opened a second time; the
second opening fails.  for example, the SvStream class uses file locking
by default.

check for code that opens the file for debug purpose and forgets to
close it, or memory leak of something that contains an open file handle.

or perhaps you have the file open in a text editor?

if it's not obvious try SysInternals Process Monitor, it can trace
system calls and their return values, and even display stack traces of
the calls.

https://technet.microsoft.com/en-us/sysinternals/bb896645.aspx


-
To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
For additional commands, e-mail: dev-h...@openoffice.apache.org