Location of documentation (Was: important info about the floatflt package and the LyX manuals)

[christian . ridderstrom]

On Tue, 5 Feb 2008, Pavel Sanda wrote:

Is there some part of the introduction or tutorial that explains that 
these documents are also available as PDFs, and where they can be 
found?


I added now a link in a note to where to find the PDFs.
(to http://wiki.lyx.org/LyX/DocumentationDevelopment)


Does this mean you added a note to one of the manuals (which)?  Or did you 
add the note to a README?



i found the wiki structure confusing.


I agree :-(


intuitively i will look for these  manuals either in
http://wiki.lyx.org/LyX/Manuals
or here
http://wiki.lyx.org/LyX/LyXHelpDocuments


Did you see the link to

http://wiki.lyx.org/LyX/Documentation

The link is there in sidebar, it is the header with the text
"LyX documentation"
but unfortunately I don't people understand it is also a link to a page 
about LyX documentation.


If you didn't notice it, I should change this somehow.


Anyway, I'd expect to find links to the PDFs from that page, i.e.

LyX/Documentation

the joke is that one find complete 1.4 documentation here. i would 
propose to somehow unite all these to some consistent state,

- i mean unity of place - only one page to be found these pdfs and unity
of version - not the mix of 1.4.5.1, 1.5.1 and 1.5.4 manuals.


Could we commit the PDFs to released branches?  In other words, when 
releasing e.g. 1.5.5, could we also generate a PDF using that version of 
LyX and then add/commit it to the repository?  If we can do this, it is 
easy to link from a wiki page (or a web page) directly to these files.


in such a case i will put link from official www pages to such kind of 
wiki download page.


If we decide to use a wiki page as a download page, we have to remember to 
give it a password. Otherwise bad people could change the download links.


A different solution for the documentation would be to generate all the 
PDFs and place them in a separate directory structure. Perhaps this could 
be on the FTP-server?


/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[Darren Freeman]

On Tue, 2008-02-05 at 08:00 +0100, Herbert Voss wrote:
> Uwe Stöhr schrieb:
> 
> >  > Generally speaking I think it sucks that the user guide doesn't 
> > compile out of the box. Have other > popular (Linux) LaTeX distributions 
> > also skipped the floatflt package?
> > 
> > TeXLive came up with the removal, others followed. For Linux TeXLive is 
> > the only active LaTeX-distribution I know.
> > Generally speaking, it sucks that they discovered that floatflt has a 
> > nosell license after 10 years and not before! Argh.
> 
> contact the author to change the license ...
> which is sometimes difficult after all these years.

This happened with prettyref as well. I looked inside prettyref when I
eventually found it, and saw how little it contained.

Surely there is somebody knowledgeable enough to re-implement floatflt
and contribute it to the distributions. Why didn't TeXLive do that
before the removal? It must have irked more than just LyX users.

Have fun,
Darren

Re: [patch] GuiSymbols

[Darren Freeman]

On Sun, 2008-02-03 at 17:09 +0100, Jürgen Spitzmüller wrote:
> from the unicodesymbols list. It's quite similar to the symbols dialogs you 
> know from the word processors (screenshot attached).

> Comments welcome.

Wow, you use LyX in black and white?

I should send you a new video card and monitor.

I'll shut up now :)

Have fun,
Darren

Re: Mathed completion popup

[rgheck]

Stefan Schimanski wrote:

Hi!

I was wondering today how complicated it would be to add a completion 
popup to mathed. Well, it took me three hours and here it is. Of 
course here and there, there are some rough edges and it has to be 
cleaned up, but it's easier than I expected.



Well, I can see how this would be pretty dang useful.

Qt is pretty amazing sometimes.

rh

Re: [Cvslog] r22802 - in /lyx-devel/trunk: lib/layouts/stdinsets.inc s...

[rgheck]

rgheck wrote:

Pavel Sanda wrote:

Author: rgheck
Date: Tue Feb  5 23:57:33 2008
New Revision: 22802

URL: http://www.lyx.org/trac/changeset/22802
Log:
Partially fix bug 4532: http://bugzilla.lyx.org/show_bug.cgi?id=4532.
See in particular comment 6.



maybe it has something in common with this.
g++ -O -o tex2lyx FloatList.o Floating.o Counters.o Layout.o 
TextClass.o Lexer.o Color.o boost.o Context.o Font.o gettext.o 
lengthcommon.o Parser.o tex2lyx.o preamble.o math.o table.o text.o  
../../src/support/.libs/liblyxsupport.a -L/usr/lib/qt4 
/usr/lib/qt4/libQtCore.so -lm -lrt -lpthread 
../../boost/.libs/liblyxboost.a /usr/lib/libaspell.so -lgpm 
/usr/lib/libSM.so /usr/lib/libICE.so -lz /usr/lib/libX11.so 
/usr/lib/libXau.so /usr/lib/libXdmcp.so -ldl
TextClass.o: In function `lyx::TextClass::read(lyx::support::FileName 
const&, lyx::TextClass::ReadType)':
TextClass.cpp:(.text+0x7cd5): undefined reference to 
`lyx::frontend::Alert::warning(std::basic_stringstd::char_traits, std::allocator > const&, 
std::basic_string, 
std::allocator > const&)'
TextClass.cpp:(.text+0x7faa): undefined reference to 
`lyx::frontend::Alert::warning(std::basic_stringstd::char_traits, std::allocator > const&, 
std::basic_string, 
std::allocator > const&)'

collect2: ld returned 1 exit status
make[2]: *** [tex2lyx] Error 1

  
Something's weird here. This is all in tex2lyx. Are you compiling with 
pch?


Oh, well, now I see the problem. I guess there must be some barrier to 
using the frontends namespace in tex2lyx. I can see why that might be, 
but it's inconvenient, to say the least.


Can someone help here?

rh

Re: [Cvslog] r22802 - in /lyx-devel/trunk: lib/layouts/stdinsets.inc s...

[Pavel Sanda]

> Author: rgheck
> Date: Tue Feb  5 23:57:33 2008
> New Revision: 22802
> 
> URL: http://www.lyx.org/trac/changeset/22802
>Log:
>Partially fix bug 4532: http://bugzilla.lyx.org/show_bug.cgi?id=4532.
>See in particular comment 6.

maybe it has something in common with this.
g++ -O -o tex2lyx FloatList.o Floating.o Counters.o Layout.o TextClass.o 
Lexer.o Color.o boost.o Context.o Font.o gettext.o lengthcommon.o Parser.o 
tex2lyx.o preamble.o math.o table.o text.o  
../../src/support/.libs/liblyxsupport.a -L/usr/lib/qt4 
/usr/lib/qt4/libQtCore.so -lm -lrt -lpthread ../../boost/.libs/liblyxboost.a 
/usr/lib/libaspell.so -lgpm /usr/lib/libSM.so /usr/lib/libICE.so -lz 
/usr/lib/libX11.so /usr/lib/libXau.so /usr/lib/libXdmcp.so -ldl
TextClass.o: In function `lyx::TextClass::read(lyx::support::FileName const&, 
lyx::TextClass::ReadType)':
TextClass.cpp:(.text+0x7cd5): undefined reference to 
`lyx::frontend::Alert::warning(std::basic_string, std::allocator > const&, 
std::basic_string, std::allocator > 
const&)'
TextClass.cpp:(.text+0x7faa): undefined reference to 
`lyx::frontend::Alert::warning(std::basic_string, std::allocator > const&, 
std::basic_string, std::allocator > 
const&)'
collect2: ld returned 1 exit status
make[2]: *** [tex2lyx] Error 1

pavel

Re: important info about the floatflt package and the LyX manuals

[Pavel Sanda]

> Yes, this should be the right place. Could you design this page the way you 
> think it should be? I'll correct then the links in the documentation to 
> this page.

Just added one section, nothing more needed IMHO - if it was on me I would
simply move the contents of download section in DocumentationDevelopment into
there. But I let it up to you.

> (Please leave http://wiki.lyx.org/LyX/DocumentationDevelopment as is.)
>
> > or here
> > http://wiki.lyx.org/LyX/LyXHelpDocuments
>
> We can set at this page a link to
> http://wiki.lyx.org/LyX/Manuals

Yes.
Pavel

Mathed completion popup

[Stefan Schimanski]

Hi!

I was wondering today how complicated it would be to add a completion  
popup to mathed. Well, it took me three hours and here it is. Of  
course here and there, there are some rough edges and it has to be  
cleaned up, but it's easier than I expected.


<>



Stefan



mathedcompletion.patch
Description: Binary data

Re: important info about the floatflt package and the LyX manuals

[Uwe Stöhr]

> i found the wiki structure confusing.

Yes, because this was originally thought as developer page.

> intuitively i will look for these manuals either in
> http://wiki.lyx.org/LyX/Manuals

Yes, this should be the right place. Could you design this page the way you think it should be? I'll 
correct then the links in the documentation to this page.

(Please leave http://wiki.lyx.org/LyX/DocumentationDevelopment as is.)

> or here
> http://wiki.lyx.org/LyX/LyXHelpDocuments

We can set at this page a link to
http://wiki.lyx.org/LyX/Manuals

regards Uwe

Re: [PATCH] Bug 2178: Empty Layout

[rgheck]

Pavel Sanda wrote:
The one remaining issue has to do with sorting, but I think there may be 
some issues with that code, anyway, since it seems to assume that the 
default layout will always appear first, and I'm not sure why that would 
have to be true.



i remember that i assumed this first position by checking the sorting of
different layouts, but you are right there is no guarantee that all layout in
the future must be done this way. its just i was not able to find counter
example in the current layouts (and also dont know what modules can bring in
this area).  
quick check revealed sorting works with your patch in case of LyXDefault.

if some problems with this appear i will sort them out (in most extreme by
stop this 'first default' property which i found to be useful.)

  

OK. We have time to sort this out. (Bad pun.)

Comments of course welcome. Better yet, fixes.



there are still missing some toolbar updates. eg open the footnote inset and
the LyXDefault appears AFTER you start typing not just after inserting the
inset and puting cursor there.  also try close document - envi toolbox remain
active and full of entries instead of disabling it. i guess switching between
tabs wouldnt update envi of the current environment too.

  
Yes, I noticed that, though it's obviously a different bug. You just 
notice it here. It's worst with ERT, where the dropbox ought to be 
disabled anyway.


Abdel?

Richard

Re: kubuntu710 and lyx

[Pavel Sanda]

> There's a more minimal debug option IIRC, was that -gfabs or something?

havent found anything like this.
pavel

Re: [PATCH] Bug 2178: Empty Layout

[Pavel Sanda]

> The one remaining issue has to do with sorting, but I think there may be 
> some issues with that code, anyway, since it seems to assume that the 
> default layout will always appear first, and I'm not sure why that would 
> have to be true.

i remember that i assumed this first position by checking the sorting of
different layouts, but you are right there is no guarantee that all layout in
the future must be done this way. its just i was not able to find counter
example in the current layouts (and also dont know what modules can bring in
this area).  
quick check revealed sorting works with your patch in case of LyXDefault.
if some problems with this appear i will sort them out (in most extreme by
stop this 'first default' property which i found to be useful.)

> Comments of course welcome. Better yet, fixes.

there are still missing some toolbar updates. eg open the footnote inset and
the LyXDefault appears AFTER you start typing not just after inserting the
inset and puting cursor there.  also try close document - envi toolbox remain
active and full of entries instead of disabling it. i guess switching between
tabs would nt update envi of the current environment too.

pavel

Re: Aussie is *slow*

[Pavel Sanda]

> > Yeah, sf would be moving out of the frying pan into the fire.  It's 
> > veeery
> > slow very often.
> 
> Maybe this is a regional issue?  sf.net has been responsive most of
> the time (US).

also few years back, but i have the same experience with the slowness, so
finally we decided to move the project from sf to our local server instead.
pavel

Re: important info about the floatflt package and the LyX manuals

[Pavel Sanda]

> > Is there some part of the introduction or tutorial that explains that 
> these documents are also
> > available as PDFs, and where they can be found?
>
> I added now a link in a note to where to find the PDFs.
> (to http://wiki.lyx.org/LyX/DocumentationDevelopment)

i found the wiki structure confusing.
intuitively i will look for these manuals either in
http://wiki.lyx.org/LyX/Manuals
or here
http://wiki.lyx.org/LyX/LyXHelpDocuments

the joke is that one find complete 1.4 documentation here.
i would propose to somehow unite all these to some consistent state,
- i mean unity of place - only one page to be found these pdfs and unity
of version - not the mix of 1.4.5.1, 1.5.1 and 1.5.4 manuals.

in such a case i will put link from official www pages to such
kind of wiki download page.

pavel

Re: Aussie is *slow*

[Joost Verburg]

Bo Peng wrote:

I used sourceforge perhaps two years ago, and it was horrible in terms of
performance at the time. Maybe they're better now, I don't know.


I have used sourceforge for three years and I do not see any reason
why lyx can not make use of it.


The performance of SF web hosting is currently much better than two 
years ago and it's definitely more reliable than the current server. 
Downloads are always very fast and a regional mirror is selected 
automatically.


And because they maintain the web/SVN/downloads servers, it saves a lot 
of time which can be spent to improve LyX. You also get nice extra 
services such as web access to SVN (similar to trac) which are 
maintained and updated by the SF people.


Joost

Re: Aussie is *slow*

[Abdelrazak Younes]

rgheck wrote:


As Andre suggested, another option would be to require forwarding, and 
then people can sort out their own spam.


Yes, I think that's the simplest and most useful solution.

Abdel.

Re: Aussie is *slow*

[Sven Hoexter]

On Mon, Feb 04, 2008 at 04:54:45PM +0100, [EMAIL PROTECTED] wrote:
> On Sat, 2 Feb 2008, Jean-Marc Lasgouttes wrote:
>
>> The machine (Athlon 2000+) has been bought from the project's fund and  
>> from time to time Lars asks on lyx-users whether somebody could renew  
>> the lyx.org name. That is all there is to it.
>
> Could we just install more memory in it?  What would it cost? (I'm  
> guessing the cost will not be that bad)


If you can find out what kind of memory you need please write me a mail.
According to another mail aussie seems to be constructed out of consumer
hardware so it's likely that we might have some spare parts in the cellar
at work to donate.

Sven
-- 
If God passed a mic to me to speak
I'd say stay in bed, world
Sleep in peace
   [The Cardigans - 03:45: No sleep]

Re: Aussie is *slow*

[rgheck]

Bernhard Roider wrote:

Pavel Sanda schrieb:

On Mon, 4 Feb 2008, rgheck wrote:

 looking at the RES memory usage, the main consumers are these. 
what if

 we start by running less childerns of spamd?

   59m   2:32.53 spamd   57m   0:52.08 spamd
   37m   1:48.72 spamd

Do we even need 'spamd'?   I mean, does it serve a useful purpose on
aussie?


i guess its purpose is to filter spam from [EMAIL PROTECTED] address and 
not doing
this filter would mean megabytes of disk space on aussie every day 
because many

people dont forward their mails etc.


shouldn't it be possible to restrict the mailbox size per user?

The worry, I think, is that the mailboxes would fill up with spam in 
that case.


As Andre suggested, another option would be to require forwarding, and 
then people can sort out their own spam.


rh

Re: important info about the floatflt package and the LyX manuals

[christian . ridderstrom]

On Tue, 5 Feb 2008, Jean-Marc Lasgouttes wrote:

note that there are many people with tex distros (tetex, older texlive 
etc) which are simply not striked by this problem, but could be striked 
by the fact the output suddenly changed from 1.5.3 to 1.5.4, so we are 
doing favour for one part of people while doing disfavour for another 
part. i dont see the gain.


The output is going to change in 1.6 anyway and we cannot avoid that. 
The move from floatflt to wrapfig was decided because of the 
shortcomings of floatflt.


I think the real "fix" should be in 1.6. Is it a real problem to do a 1.6 
that in practice only contains this as the major difference?


/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[Uwe Stöhr]

> Is there some part of the introduction or tutorial that explains that these 
documents are also
> available as PDFs, and where they can be found?

I added now a link in a note to where to find the PDFs.
(to http://wiki.lyx.org/LyX/DocumentationDevelopment)

regards Uwe

Re: Aussie is *slow*

[Bernhard Roider]

Pavel Sanda schrieb:

On Mon, 4 Feb 2008, rgheck wrote:


 looking at the RES memory usage, the main consumers are these. what if
 we start by running less childerns of spamd?

   59m   2:32.53 spamd   57m   0:52.08 spamd
   37m   1:48.72 spamd

Do we even need 'spamd'?   I mean, does it serve a useful purpose on
aussie?


i guess its purpose is to filter spam from [EMAIL PROTECTED] address and not 
doing
this filter would mean megabytes of disk space on aussie every day because many
people dont forward their mails etc.


shouldn't it be possible to restrict the mailbox size per user?

bernhard



We should check on which addresses it is run. I suspect there is one
instance per mail address...


afaik spamd has fixed number of childs (here 3); so its just to change this
parameter in config.

pavel

Re: Aussie is *slow*

[Bo Peng]

> I used sourceforge perhaps two years ago, and it was horrible in terms of
> performance at the time. Maybe they're better now, I don't know.


I have used sourceforge for three years and I do not see any reason
why lyx can not make use of it.

Bo

Re: important info about the floatflt package and the LyX manuals

[Herbert Voss]

rgheck schrieb:

Herbert Voss wrote:

Uwe Stöhr schrieb:
 > I think we should not include floatflt in the Windows installer. 
This would mean that the

 > restrictive floatflt license conditions also apply to the installer.

But that is no problem, because we don't want donations for the 
installer.


that is not the point ...
None of the TeX users group get money for selling the TeX-Collection.
But it is not allowed to have "nosell" packages on the DVD, when
book stores sell the DVD or publishers put the DVD for free in
a journal. You have to pay for the journal together with the DVD,
hence it is not free.

So if the Windows installer were included on such a DVD, that would 
violate the terms of the floatflt license. That's the point, yes?


yes

Herbert

Re: important info about the floatflt package and the LyX manuals

[rgheck]

Herbert Voss wrote:

Uwe Stöhr schrieb:
 > I think we should not include floatflt in the Windows installer. 
This would mean that the

 > restrictive floatflt license conditions also apply to the installer.

But that is no problem, because we don't want donations for the 
installer.


that is not the point ...
None of the TeX users group get money for selling the TeX-Collection.
But it is not allowed to have "nosell" packages on the DVD, when
book stores sell the DVD or publishers put the DVD for free in
a journal. You have to pay for the journal together with the DVD,
hence it is not free.

So if the Windows installer were included on such a DVD, that would 
violate the terms of the floatflt license. That's the point, yes?


rh

Re: important info about the floatflt package and the LyX manuals

[Herbert Voss]

Uwe Stöhr schrieb:
 > I think we should not include floatflt in the Windows installer. This 
would mean that the

 > restrictive floatflt license conditions also apply to the installer.

But that is no problem, because we don't want donations for the installer.


that is not the point ...
None of the TeX users group get money for selling the TeX-Collection.
But it is not allowed to have "nosell" packages on the DVD, when
book stores sell the DVD or publishers put the DVD for free in
a journal. You have to pay for the journal together with the DVD,
hence it is not free.

Herbert

Re: Aussie is *slow*

[Andre Poenitz]

On Tue, Feb 05, 2008 at 10:44:59AM +0100, Pavel Sanda wrote:
> > > On Mon, 4 Feb 2008, rgheck wrote:
> > >
> > >>>  looking at the RES memory usage, the main consumers are these.
> > >>>  what if we start by running less childerns of spamd?
> > >>>
> > >>>59m   2:32.53 spamd   57m   0:52.08
> > >>>spamd 37m   1:48.72 spamd
> > >
> > > Do we even need 'spamd'?   I mean, does it serve a useful purpose
> > > on aussie?
> 
> i guess its purpose is to filter spam from [EMAIL PROTECTED] address and
> not doing this filter would mean megabytes of disk space on aussie
> every day because many people dont forward their mails etc.

We should just enforce mail forwarding.

Andre'

Re: important info about the floatflt package and the LyX manuals

[Helge Hafting]

Pavel Sanda wrote:

Uwe Stöhr wrote:

Unfortunately using wrapfig like in trunk cannot be backported since it is 
a fileformat change.
  

Well - this rule doesn't have to be absolute!



this is not just about the rules. being as user used to increace from x to
1.y.x+1 without any worry about _the latex output_ is very important imho. 
i will not study way too much release notes for such an update. but i will

enquire what this 1.6 change mean, perfectly knowing i could face some problems
otherwise.
  

I see.  So the only option is "wait for 1.6", possibly with
a release of 1.5.4 renamed 1.6 with this fix.

Helge Hafting

Re: Aussie is *slow*

[Uwe Stöhr]

> What about http://www.berlios.de/ ? I used that long time ago when they 
already had subversion,
> and SF did not, maybe 2 years ago. Always felt fast at that time.

Me not. Since 3 years I use berlios.de for the LyXWinInstaller and when you have a problem nobody 
every replies on support requests. For example is the ftp-server upload broken since 2 years and 
it's definively their problem. And berlios.de is often down.


regards Uwe

Re: proposal to upgrade Aussie

[christian . ridderstrom]

On Tue, 5 Feb 2008, Lars Gullik Bjønnes wrote:


So the questions would then be:
* Can we upgrade the memory?


Yes. I have another box hat I don't use anymore (story about water 
cooling and leakage) with 1GB RAM that could be installed into Aussie.


Ah... always good with "extra" information, think I'll stay away from 
water cooling on my next machine then.



* What kind of RAM does it take?
* What kind of HDD could we add to it?


IDE

There is a mirror setup currently, but I don't trust it anymore.

I am not sure that we really need more disk, but the partitioning needs 
to change. And the disks are old, so they should be changed as well...


Size of disks from 'df -h' (only the large parts):

FilesystemSize  Used Avail Use% Mounted on
/dev/md21G  313M  644M  33% /
/dev/md4   61G   25G   33G  44% /home
/dev/md1  9.7G  2.2G  7.0G  25% /usr
/dev/md3  2.0G  1.7G  265M  87% /var
   ==  74G

So currently the disk(s) are roughly 2 x 74 GB with about (2x) 25GB free.

I checked on dustin.se, and we could get 2 x 80 GB for about 70 EUR. I'm 
sure we could get old HDDs donated to the project, but it's probably safer 
to get brand new HDDs instead. If the fund doesn't cover this, I'll donate 
the money.  Normally I'd get larger disks when upgrading, but I guess that 
with 25 GB available, we really don't need more space.



* Is it practially possible for us to do this? Does anyone have physical
   access?


That is the grit of it... I have physical access if I want. Just have to 
make an appointment with trolltech. I havent seen the box since before 
they changed offices.


Do you have the time/energy/willingness to do this?   If the answer is 
yes, within what time frame?


Best regards (and nice to hear from you again!)
/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

[PATCH] Bug 2178: Empty Layout

[rgheck]

The attached patch is now pretty much complete, I think, so I'm sending 
it again to the list.


The basic idea has two parts. First, we hard code an "empty layout"
(called LyXDefault, for want of a better name) in TextClass and read it
before doing anything else. It can therefore be customized by classes,
if they want---say, to make it left-aligned. Second, InsetText's are
divided into three types: (i) normal ones, that use the "default" layout
defined by the text class; (ii) highly restrictive ones, such as ERT and
(not quite an inset) table cells, which demand the empty layout; (iii)
middling ones, which default to an empty layout and use the empty layout
in place of the default. (This is so we don't get the same problem we
had with ERT in e.g. footnotes.) The type of inset is signaled by new
methods InsetText::forceEmptyLayout() and InsetText::useEmptyLayout().
(The latter might better be called: useEmptyLayoutInsteadOfDefault(),
but that's silly.) The old InsetText::forceDefaultParagraphs() has been
split into these, plus a new method
InsetText::allowParagraphCustomization(). A lot of the changes just
adapt to this change.

The other big change is in GuiToolbar: We want to show LyXDefault and
the "default" layout only when they're active.

The one remaining issue has to do with sorting, but I think there may be 
some issues with that code, anyway, since it seems to assume that the 
default layout will always appear first, and I'm not sure why that would 
have to be true.


Comments of course welcome. Better yet, fixes.

rh


Index: src/TextClass.h
===
--- src/TextClass.h	(revision 22792)
+++ src/TextClass.h	(working copy)
@@ -109,7 +109,14 @@
 	docstring const & defaultLayoutName() const;
 	///
 	LayoutPtr const & defaultLayout() const;
+	/// returns a special layout for use when we don't really want one,
+	/// e.g., in table cells
+	LayoutPtr const & emptyLayout() const 
+			{ return operator[](emptylayout_); };
 	///
+	docstring const & emptyLayoutName() const 
+			{ return emptylayout_; }
+	///
 	std::string const & name() const;
 	///
 	docstring const & labelstring() const;
@@ -200,6 +207,8 @@
 	std::string class_header_;
 	///
 	docstring defaultlayout_;
+	/// name of empty layout
+	static const docstring emptylayout_;
 	/// preamble text to support layout styles
 	docstring preamble_;
 	/// latex packages loaded by document class.
Index: src/TextClass.cpp
===
--- src/TextClass.cpp	(revision 22792)
+++ src/TextClass.cpp	(working copy)
@@ -34,6 +34,7 @@
 #include "support/os.h"
 
 #include 
+#include 
 
 using namespace std;
 using namespace lyx::support;
@@ -111,6 +112,9 @@
 }
 
 
+docstring const TextClass::emptylayout_ = from_ascii("LyXDefault");
+
+
 bool TextClass::isTeXClassAvailable() const
 {
 	return texClassAvail_;
@@ -214,6 +218,27 @@
 	}
 	LYXERR(Debug::TCLASS, to_utf8(makeDisplayPath(filename.absFilename(;
 
+	// Define the `empty' layout used in table cells, ert, etc. Note that 
+	// we do this before loading any layout file, so that classes can 
+	// override features of this layout if they should choose to do so.
+	if (rt == BASECLASS) {
+		static char const * s = "Margin Static\n"
+			"LatexType Paragraph\n"
+			"LatexName dummy\n"
+			"Align Block\n"
+			"AlignPossible Left, Right, Center\n"
+			"LabelType No_Label\n"
+			"End";
+		istrstream ss(s);
+		Lexer lex(textClassTags, sizeof(textClassTags) / sizeof(textClassTags[0]));
+		lex.setStream(ss);
+		Layout lay;
+		lay.setName(emptylayout_);
+		bool error = readStyle(lex, lay);
+		BOOST_ASSERT(!error);
+		layoutlist_.push_back(boost::shared_ptr(new Layout(lay)));
+	}
+
 	Lexer lexrc(textClassTags,
 		sizeof(textClassTags) / sizeof(textClassTags[0]));
 
@@ -483,7 +508,6 @@
 		}
 		LYXERR(Debug::TCLASS, "Minimum TocLevel is " << min_toclevel_
 			<< ", maximum is " << max_toclevel_);
-
 	}
 
 	return error;
@@ -1028,7 +1052,7 @@
 
 bool TextClass::deleteLayout(docstring const & name)
 {
-	if (name == defaultLayoutName())
+	if (name == defaultLayoutName() || name == emptyLayoutName())
 		return false;
 
 	LayoutList::iterator it =
Index: src/frontends/qt4/GuiToolbar.h
===
--- src/frontends/qt4/GuiToolbar.h	(revision 22792)
+++ src/frontends/qt4/GuiToolbar.h	(working copy)
@@ -24,6 +24,7 @@
 
 namespace lyx {
 
+class Inset;
 class TextClass;
 class ToolbarItem;
 
@@ -53,6 +54,7 @@
 private:
 	GuiView & owner_;
 	TextClass const * text_class_;
+	Inset const * inset_;
 };
 
 
Index: src/frontends/qt4/GuiToolbar.cpp
===
--- src/frontends/qt4/GuiToolbar.cpp	(revision 22792)
+++ src/frontends/qt4/GuiToolbar.cpp	(working copy)
@@ -304,16 +304,23 @@
 		clear();
 		setEnabled(false);
 		text_class_ = 0;
+		inset_ = 0;
 		return;
 	}
 
 	setEnabled(true);
 	TextClass const * text_cl

Re: important info about the floatflt package and the LyX manuals

[Uwe Stöhr]

> I think we should not include floatflt in the Windows installer. This would 
mean that the
> restrictive floatflt license conditions also apply to the installer.

But that is no problem, because we don't want donations for the installer.

regards Uwe

Re: bugs needed to be fixed to make trunk usable

[rgheck]

Transferring this discussion to the list.

http://bugzilla.lyx.org/show_bug.cgi?id=4532

/--- Additional Comment #3 
 From Richard Heck 
 2008-02-04 20:37 --- /


We should consider hard-coding the loading of
stdinsets.inc in TextClass.cpp. stdinsets.inc isn't like other included files.
This is basic LyX stuff. It's just in an include file because, well, that's
where it is. So, minimally, you could do roughly this:
 if (rt == BASECLASS) {
   FileName tmp = libFileSearch("layouts", "stdinsets.inc");
   if (tmp.empty()) 
 //errormsg

   else {
 if (read(tmp, MERGE)) {
   //errormsg
 }
  }
This could be loaded early, before anything is actually read, so that users
could override stdinsets.inc.


/--- Additional Comment #6 
 From JMarc 
 2008-02-05 10:13 --- /


We could also load stdinsets.inc after the textclass if no insets have been
defined, as a fallback. Or even add a "Provides insetsdefs" in stdinsets.inc and
load stdinsets.inc if the insetdefs feature has not been provided yet.

However, I believe we need to include it properly in all textclasses we 
distribute.


/--- Additional Comment #7 
 From Richard Heck 
 2008-02-05 17:26 --- /


That last suggestion sounds fine by me. But I don't understand the Provides
stuff, so I'll have to leave that to someone else. Fixing our own layouts is
just a matter of doing the grunt work.

By the way, it occurred to me that there is another problem here. We get the
invalid Flex inset because an "undefined" inset doesn't get written properly.
You could therefore get a similar problem if you tried to save a file with an
inset that was of undefined type, no matter why: E.g., if someone sent you a
file that used a module you didn't have. Instead, it should be written as an
inset of type "undefined", where that type of inset is hard-coded somewhere.

Re: changeset 22764 also for branch

[Uwe Stöhr]

> The beamer class is not really a model for good layout implementation.

Maybe, but that letters can have parts behind the \end{letter} command is a fact. And the "bad" 
implementation of beamer is a consequence of missing LyX support for certain commands.


> I do not remember exactly why I put in the hack back then,

Because when \end{letter} is always the end, it works and people don't have to take care about this. 
I guess your added the hack to the adress, because without an address, a letter is senseless.


> but does changing the LaTeXType to Environment do the trick?

I don't understand what you mean. You can check out if your idea works by using the serial letter 
example files in trunk.


Attached is what I want to have in branch. You can see that serial letters work also fine with 
branch - when you start with the example files. So can I proceed?


regards Uwe
\adrentry{Boehme}
 {Wilhelm}
 {Ubierweg 49\\754127 Dingsda}
 {0227557/35448}
 {m}
 {}
 {alter Kollege}
 {WILBOE}
\adrentry{Zylla}
 {Rieke}
 {Salierweg 9\\850289 Kolonia}
 {022811/458290}
 {w}
 {}
 {Versicherung}
 {RIEZY}
\adrentry{Klausen}
 {Daniela}
 {Promenadenweg 17\\23155 Fuhlingen}
 {0672289/30945}
 {w}
 {}
 {Freundin}
 {DANKLA}
\adrentry{Tollmann}
 {Gerd}
 {Ginsterweg 119\\48890 Karlingen}
 {074418/90833}
 {m}
 {}
 {Urlaub}
 {GERTO}


serienbrief1.lyx
Description: application/lyx

Re: Aussie is *slow*

[christian . ridderstrom]

On Tue, 5 Feb 2008, Bo Peng wrote:


I used sourceforge perhaps two years ago, and it was horrible in terms of
performance at the time. Maybe they're better now, I don't know.


I have used sourceforge for three years and I do not see any reason why 
lyx can not make use of it.


One reason is the lack of backup of wiki data with the simple use of 
PmWiki. To be more specific, the project web space is mounted read-only, 
meaning any directory that may be served directly by SourceForge.net is 
not writable by the web server. In practice, using a flat filesystem as 
database, the data has to be stored in /tmp/persistent...


This problem can be mitigated with manual backups that we'd do on a 
frequent basis though.


/C

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[Uwe Stöhr]

> Is there really a lyx2lyx entry related to this change?

I checked my commit history. When we only use a very basic implementation, I 
think we could backport it:
http://www.lyx.org/trac/changeset/20185

This was the first commit that changes wrap floats from floatflt to wrapfig. It only need a dummy 
fileformat change since the content of the LyX-files is not really changed, only the functionality 
in the background.


For a proper wrapfig support we need fileformat changes:
http://www.lyx.org/trac/changeset/20462
http://www.lyx.org/trac/changeset/20713

But with the basic support, we have the same state of functionality as with 
floatflt.

So I think we can backport it. What do you think? Jürgen, your opionion?

regards Uwe

Re: [Cvslog] r22782 - in /lyx-devel/trunk/src: CutAndPaste.cpp factory...

[rgheck]

Thanks for sorting this out, Stefan. I hadn't looked at this bug. But 
let me urge everyone to look at this patch. It reflects a change in 
TextClass handling that is important to understand.


The point, as Stefan says, is that anything that belongs to a text class 
needs to be "preserved" by keeping a TextClassPtr around. These are 
boost:shared_ptr's. The need for this is a consequence of the modules 
code. TextClass objects do not now represent layout files but rather the 
result of incorporating whatever modules the user loads. So they are 
unique to documents and can even change if the user loads a new module. 
So you cannot just represent a TextClass by an index into a list of 
layout files, as we used to. Rather, a TextClass is just an object, and 
it is inclined to disappear when the document that created it 
disappears. Now the problem is that, if you cut something, the cut 
material needs to know about its TextClass, since that is what defines 
its layout, etc. So the TextClass needs to "stay with" what's cut, since 
the document itself may be closed. And all references to layout-related 
stuff in, say, an inset need to go via the corresponding TextClassPtr.


There may be an argument here for doing things differently, a way Abdel 
suggested a while ago, namely, that we keep every TextClass we ever 
generate in (say) a std::vector and refer to them via indices into this 
vector. This change would not be terribly hard to make: It's just a 
matter of redefining TextClassPtr. The downside to this is that you 
could end up with a lot of TextClass's in memory that don't need to be 
there. But it would prevent this kind of problem.


Richard

[EMAIL PROTECTED] wrote:

Author: sts
Date: Tue Feb  5 11:34:01 2008
New Revision: 22782

URL: http://www.lyx.org/trac/changeset/22782
Log:
* Do not keep pointers to data structures around if you don't know
  whether the data structure outlives the pointer:

  InsetLayouts are really owned by the corresponding TextClass. So
  keep the TextClass alive by keeping a TextClassPtr around for the
  pointers lifetime. This fixes Bug #4538.

Modified:
lyx-devel/trunk/src/CutAndPaste.cpp
lyx-devel/trunk/src/factory.cpp
lyx-devel/trunk/src/insets/InsetCollapsable.cpp
lyx-devel/trunk/src/insets/InsetCollapsable.h
lyx-devel/trunk/src/insets/InsetFlex.cpp
lyx-devel/trunk/src/insets/InsetFlex.h

Modified: lyx-devel/trunk/src/CutAndPaste.cpp
URL: http://www.lyx.org/trac/file/lyx-devel/trunk/src/CutAndPaste.cpp?rev=22782
==
--- lyx-devel/trunk/src/CutAndPaste.cpp (original)
+++ lyx-devel/trunk/src/CutAndPaste.cpp Tue Feb  5 11:34:01 2008
@@ -445,17 +445,15 @@
if (inset->lyxCode() != FLEX_CODE)
// FIXME: Should we verify all InsetCollapsable?
continue;
-   docstring const name = inset->name();
-   InsetLayout const & il = tclass2.insetlayout(name);
-   inset->setLayout(il);
-   if (il.labelstring != from_utf8("UNDEFINED"))
+   inset->setLayout(c2);
+   if (inset->getLayout().labelstring != from_utf8("UNDEFINED"))
continue;
// The flex inset is undefined in tclass2
docstring const s = bformat(_(
"Flex inset %1$s is "
"undefined because of class "
"conversion from\n%2$s to %3$s"),
-   name, from_utf8(tclass1.name()),
+   inset->name(), from_utf8(tclass1.name()),
from_utf8(tclass2.name()));
// To warn the user that something had to be done.
errorlist.push_back(ErrorItem(

Modified: lyx-devel/trunk/src/factory.cpp
URL: http://www.lyx.org/trac/file/lyx-devel/trunk/src/factory.cpp?rev=22782
==
--- lyx-devel/trunk/src/factory.cpp (original)
+++ lyx-devel/trunk/src/factory.cpp Tue Feb  5 11:34:01 2008
@@ -102,9 +102,7 @@
 
 		case LFUN_FLEX_INSERT: {

string s = cmd.getArg(0);
-   TextClass const & tclass = params.getTextClass();
-   InsetLayout const & il = 
tclass.insetlayout(from_utf8(s));
-   return new InsetFlex(params, il);
+   return new InsetFlex(params, params.getTextClassPtr(), 
s);
}
 
 		case LFUN_NOTE_INSERT: {

@@ -384,8 +382,6 @@
}
 
 	auto_ptr inset;

-
-   TextClass const & tclass = buf.params().getTextClass();
 
 	lex.next();

string tmptok = lex.getString();
@@ -480,8 +476,8 @@
} else if (tmptok == "Flex") {
lex.next();
string s = lex.getString();
-   InsetLayout const & il = 
tclass.insetlayout(from_utf8(s));
- 

Re: Aussie is *slow*

[Bo Peng]

> Yeah, sf would be moving out of the frying pan into the fire.  It's veeery
> slow very often.

Maybe this is a regional issue?  sf.net has been responsive most of
the time (US).

Bo

Re: important info about the floatflt package and the LyX manuals

[Uwe Stöhr]

> contact the author to change the license ...
> which is sometimes difficult after all these years.

I already did this, but got no reply until now.
If I get no reply till the end of this week, I'll try to phone him.

regards Uwe

Re: Aussie is *slow*

[Pavel Sanda]

> > vss should be complete mem (shared and swapped) rss should be resident mem
> > (without swap).
> > spamd and httpd are sums of all childerns (plus one grep instance;)
> 
> But part of this memory is shared between the children, isn't it?

not sure:) but i believe most of mem is used by some tokens data individual for 
each spamd, while the actual shared code and libs will be rather small.
there was some flag for ps to show shared memory iirc.

however the main point was to find correlation between black curve and other
curves. now i see we have 3 spamd, so lets wait what will happen when
some swap crisis will come again.

pavel

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

Pavel Sanda <[EMAIL PROTECTED]> writes:

> vss should be complete mem (shared and swapped) rss should be resident mem
> (without swap).
> spamd and httpd are sums of all childerns (plus one grep instance;)

But part of this memory is shared between the children, isn't it?

>> Looks like it tries to update its configuration and fails. Shall I
>> just uninstall pyzor?
>
> dont know. anyway i dont believe this has anything to do with performance
> problems.

I fixed the server address for pyzor, and do not see zombie processes
(yet).

JMarc

Re: Aussie is *slow*

[Pavel Sanda]

> By looking at ps as a tree, I notice that the spamd children can have
> a  pyzor child. You cannot see that right now, because I
> restarted spamd.

i established small script which checks the aussie system info every 5 min.
the result is updated here 
http://195.113.31.123/~sanda/junk/aussie_load.png
from time to time (i tried to put it in my page on aussie, but it seems i have
no privileges to display ~/sanda in apache conf?).

note carefuly how the load came down from in the begining, when i start
the measurement - it was in the time you have 4 spamd childern launched
on aussie then you put it back to 3 instances.

vss should be complete mem (shared and swapped) rss should be resident mem
(without swap).
spamd and httpd are sums of all childerns (plus one grep instance;)
pavel

> I tried to see where pyzor is invoked in spamd configuration, but
> failed. However, it turns out that pyzor is broken: [EMAIL PROTECTED]
> .spamassassin]# pyzor check  servers from http://pyzor.sourceforge.net/cgi-bin/inform-servers-0-3-x
> Traceback (most recent call last): File "/usr/bin/pyzor", line 4, in ?
> pyzor.client.run() File
> "/usr/lib/python2.4/site-packages/pyzor/client.py", line 934, in run
> ExecCall().run() File
> "/usr/lib/python2.4/site-packages/pyzor/client.py", line 175, in run
> download(config.get('client', 'DiscoverServersURL'), servers_fn) File
> "/usr/lib/python2.4/site-packages/pyzor/client.py", line 943, in
> download urllib.urlretrieve(url, outfile) File
> "/usr/lib/python2.4/urllib.py", line 89, in urlretrieve return
> _urlopener.retrieve(url, filename, reporthook, data) File
> "/usr/lib/python2.4/urllib.py", line 222, in retrieve fp =
> self.open(url, data) File "/usr/lib/python2.4/urllib.py", line 190, in
> open return getattr(self, name)(url) File
> "/usr/lib/python2.4/urllib.py", line 313, in open_http h.endheaders()
> File "/usr/lib/python2.4/httplib.py", line 798, in endheaders
> self._send_output() File "/usr/lib/python2.4/httplib.py", line 679, in
> _send_output self.send(msg) File "/usr/lib/python2.4/httplib.py", line
> 646, in send self.connect() File "/usr/lib/python2.4/httplib.py", line
> 614, in connect socket.SOCK_STREAM): IOError: [Errno socket error] (2,
> 'No such file or directory')
> 
> 
> Looks like it tries to update its configuration and fails. Shall I
> just uninstall pyzor?

dont know. anyway i dont believe this has anything to do with performance
problems.

pavel

> JMarc

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

Pavel Sanda <[EMAIL PROTECTED]> writes:

>> update: i looked on spamd now - one of our restarted 30mb spamd is again on 
>> 64 mb
>> in 5 mins :(
>
> and update after 10mins:
> 165692 32104
> 176652 76524
> 31124  18768
>
> so in total memory we are on +270mb in 10 minutes.

By looking at ps as a tree, I notice that the spamd children can have
a  pyzor child. You cannot see that right now, because I
restarted spamd.


I tried to see where pyzor is invoked in spamd configuration, but
failed. However, it turns out that pyzor is broken: [EMAIL PROTECTED]
.spamassassin]# pyzor check http://pyzor.sourceforge.net/cgi-bin/inform-servers-0-3-x
Traceback (most recent call last): File "/usr/bin/pyzor", line 4, in ?
pyzor.client.run() File
"/usr/lib/python2.4/site-packages/pyzor/client.py", line 934, in run
ExecCall().run() File
"/usr/lib/python2.4/site-packages/pyzor/client.py", line 175, in run
download(config.get('client', 'DiscoverServersURL'), servers_fn) File
"/usr/lib/python2.4/site-packages/pyzor/client.py", line 943, in
download urllib.urlretrieve(url, outfile) File
"/usr/lib/python2.4/urllib.py", line 89, in urlretrieve return
_urlopener.retrieve(url, filename, reporthook, data) File
"/usr/lib/python2.4/urllib.py", line 222, in retrieve fp =
self.open(url, data) File "/usr/lib/python2.4/urllib.py", line 190, in
open return getattr(self, name)(url) File
"/usr/lib/python2.4/urllib.py", line 313, in open_http h.endheaders()
File "/usr/lib/python2.4/httplib.py", line 798, in endheaders
self._send_output() File "/usr/lib/python2.4/httplib.py", line 679, in
_send_output self.send(msg) File "/usr/lib/python2.4/httplib.py", line
646, in send self.connect() File "/usr/lib/python2.4/httplib.py", line
614, in connect socket.SOCK_STREAM): IOError: [Errno socket error] (2,
'No such file or directory')


Looks like it tries to update its configuration and fails. Shall I
just uninstall pyzor?

JMarc

Re: Aussie is *slow*

[Stefan Schimanski]

Am 05.02.2008 um 14:52 schrieb Hans Meine:

Am Montag, 04. Februar 2008 16:50:55 schrieb [EMAIL PROTECTED] 
:

On Sun, 3 Feb 2008, Joost Verburg wrote:
Maybe it would be better to switch to SourceForge or something  
similar.
They provide web hosting, SVN, mailing lists, download mirrors  
etc., and

it's much more reliable than the current web / FTP servers.


I used sourceforge perhaps two years ago, and it was horrible in  
terms of

performance at the time. Maybe they're better now, I don't know.


Yeah, sf would be moving out of the frying pan into the fire.  It's  
veeery

slow very often.


What about http://www.berlios.de/ ? I used that long time ago when  
they already had subversion, and SF did not, maybe 2 years ago. Always  
felt fast at that time.


Stefan

Re: Aussie is *slow*

[Abdelrazak Younes]

Hans Meine wrote:

Am Montag, 04. Februar 2008 16:50:55 schrieb [EMAIL PROTECTED]:

On Sun, 3 Feb 2008, Joost Verburg wrote:

Maybe it would be better to switch to SourceForge or something similar.
They provide web hosting, SVN, mailing lists, download mirrors etc., and
it's much more reliable than the current web / FTP servers.

I used sourceforge perhaps two years ago, and it was horrible in terms of
performance at the time. Maybe they're better now, I don't know.


Yeah, sf would be moving out of the frying pan into the fire.  It's veeery 
slow very often.


That's a valid description of aussie too ;-)

Re: Aussie is *slow*

[Hans Meine]

Am Montag, 04. Februar 2008 16:50:55 schrieb [EMAIL PROTECTED]:
> On Sun, 3 Feb 2008, Joost Verburg wrote:
> > Maybe it would be better to switch to SourceForge or something similar.
> > They provide web hosting, SVN, mailing lists, download mirrors etc., and
> > it's much more reliable than the current web / FTP servers.
>
> I used sourceforge perhaps two years ago, and it was horrible in terms of
> performance at the time. Maybe they're better now, I don't know.

Yeah, sf would be moving out of the frying pan into the fire.  It's veeery 
slow very often.

-- 
Ciao, /  /
 /--/
/  / ANS

Re: important info about the floatflt package and the LyX manuals

[Jean-Marc Lasgouttes]

Pavel Sanda <[EMAIL PROTECTED]> writes:

> note that there are many people with tex distros (tetex, older texlive etc)
> which are simply not striked by this problem, but could be striked
> by the fact the output suddenly changed from 1.5.3 to 1.5.4, so we are doing
> favour for one part of people while doing disfavour for another
> part. i dont see the gain. 

The output is going to change in 1.6 anyway and we cannot avoid that.
The move from floatflt to wrapfig was decided because of the
shortcomings of floatflt.

Of course, documents using wrapped floats are fragile beasts, and the
change of package will probably perturb the output.

JMarc

Re: important info about the floatflt package and the LyX manuals

[Pavel Sanda]

> I think even the dummy file format change was not needed, since there
> is nothing lyx2lyx can do to alleviate whatever problem may appear.
> This change of package may change the output, there not much we can do
> about it.
> 
> I propose to backport a subset of this (even the GUI changes may not
> be necessary).

note that there are many people with tex distros (tetex, older texlive etc)
which are simply not striked by this problem, but could be striked by the fact
the output suddenly changed from 1.5.3 to 1.5.4, so we are doing favour for one
part of people while doing disfavour for another part. i dont see the gain.

pavel

Re: Aussie is *slow*

[Pavel Sanda]

> update: i looked on spamd now - one of our restarted 30mb spamd is again on 
> 64 mb
> in 5 mins :(

and update after 10mins:
165692 32104
176652 76524
31124  18768

so in total memory we are on +270mb in 10 minutes.
p

Re: Aussie is *slow*

[Pavel Sanda]

> Let's see first what happens now that I have restarted. It may be that
> 163M for one of the process was some kind of accident. 

i have followed aussie top for some time and my experience is that
this size dynamically changes in both directions. after all look on
my mail from few days back - 59, 57, 37. today it was 108, 21, 22.


beeing in this debate, i look also to httpd processes - before 20 mins
root 23768  0.0  0.8  21408  2160 ?Ss   Feb04   0:02 /usr/sbin/httpd
apache   20698  2.6  7.9  37140 20264 ?R12:36   0:05 /usr/sbin/httpd
apache   20714  4.2  6.7  34860 17312 ?S12:36   0:07 /usr/sbin/httpd
apache   20805  8.8  7.4  36532 18936 ?R12:39   0:03 /usr/sbin/httpd
apache   20818  5.2  7.8  37336 20136 ?S12:39   0:01 /usr/sbin/httpd

now:
apache   21096  0.8  4.7  34976 12232 ?S12:49   0:03 /usr/sbin/httpd
apache   21121  1.2  3.5  28804  9116 ?S12:50   0:04 /usr/sbin/httpd
apache   21138  0.2  1.9  30020  4972 ?S12:50   0:00 /usr/sbin/httpd
apache   21140  1.3  5.8  34428 14876 ?D12:50   0:04 /usr/sbin/httpd
apache   21144  1.6  5.6  34328 14428 ?D12:51   0:04 /usr/sbin/httpd
apache   21151  1.0  5.9  33956 15128 ?D12:52   0:01 /usr/sbin/httpd
apache   21153  0.4  3.2  30012  8200 ?D12:52   0:00 /usr/sbin/httpd
apache   21162  0.7  3.2  28836  8284 ?D12:53   0:00 /usr/sbin/httpd
apache   21165  2.2  3.5  28840  9148 ?D12:53   0:02 /usr/sbin/httpd
apache   21186  0.6  4.4  28948 11380 ?D12:54   0:00 /usr/sbin/httpd
apache   21193  1.5  5.0  32924 13028 ?D12:54   0:01 /usr/sbin/httpd
apache   21199  0.7  3.1  27860  8152 ?D12:54   0:00 /usr/sbin/httpd
apache   21210  0.8  3.1  27596  7980 ?D12:55   0:00 /usr/sbin/httpd
apache   21214  4.8  3.2  28020  8280 ?D12:55   0:01 /usr/sbin/httpd
apache   21215  0.0  1.5  21772  3988 ?S12:55   0:00 /usr/sbin/httpd
apache   21216  0.2  2.4  26136  6200 ?D12:55   0:00 /usr/sbin/httpd
apache   21217  0.1  2.0  21720  5312 ?D12:55   0:00 /usr/sbin/httpd
apache   21219  0.0  0.5  21408  1412 ?S12:55   0:00 /usr/sbin/httpd

cant be this forking also somehow restricted?


update: i looked on spamd now - one of our restarted 30mb spamd is again on 64 
mb
in 5 mins :(

pavel

Re: r22779 - in /lyx-devel/trunk/src/frontends/qt4: GuiWorkAr...

[Abdelrazak Younes]

Stefan Schimanski wrote:


void GuiWorkArea::processKeySym(KeySymbol const & key, KeyModifier mod)
{
// In order to avoid bad surprise in the middle of an operation,
-// we better stop the blinking cursor.
+// we better stop the blinking cursor...
stopBlinkingCursor();
+// ...and the general timer.
+stopGeneralTimer();

theLyXFunc().setLyXView(lyx_view_);
theLyXFunc().processKeySym(key, mod);
+   
+// the cursor gets restarted in GuiView::restartCursor()

+startGeneralTimer();
}



This very last call is a bad idea. LFUN can kill the workarea when you 
close the buffer via the keyboard. So we are inside of a method of an 
deallocated object.


Right. The best solution is I think to transfer the general timer to 
GuiView or GuiApplication.


Abdel.

Re: r22779 - in /lyx-devel/trunk/src/frontends/qt4: GuiWorkAr...

[Abdelrazak Younes]

Stefan Schimanski wrote:

Now I get a segfault, when closing a buffer via the keyboard shortcut:


Oh shxxt, this problem was hidden somewhere in the back of my mind but 
didn't manage to come out fully ;-/


This is because the key kills the Buffer which in turn kills the 
WorkArea. So the timer does not exist anymore...


One solution would be to just transfer the general timer to 
GuiApplication. Then, there is no risk of disappearance.


Abdel.



#0  0x024c7197 in QObject::startTimer ()
#1  0x024d0381 in QTimer::start ()
#2  0x0048be48 in lyx::frontend::GuiWorkArea::startGeneralTimer 
(this=0x1a25a030) at qt4/GuiWorkArea.h:125
#3  0x00486e41 in lyx::frontend::GuiWorkArea::processKeySym 
(this=0x1a25a030, [EMAIL PROTECTED], mod=lyx::ControlModifier) at 
/Users/sts/Quellen/mac/lyx-scope-dynmacro/src/frontends/qt4/GuiWorkArea.cpp:345 

#4  0x00489376 in lyx::frontend::GuiWorkArea::keyPressEvent 
(this=0x1a25a030, ev=0xbfffe970) at 
/Users/sts/Quellen/mac/lyx-scope-dynmacro/src/frontends/qt4/GuiWorkArea.cpp:737 


#5  0x01ac1a48 in QWidget::event ()
#6  0x01d70eb1 in QFrame::event ()
#7  0x01df42f6 in QAbstractScrollArea::event ()
#8  0x00489568 in lyx::frontend::GuiWorkArea::event (this=0x1a25a030, 
e=0xbfffe970) at 
/Users/sts/Quellen/mac/lyx-scope-dynmacro/src/frontends/qt4/GuiWorkArea.cpp:544 



Stefan

Am 05.02.2008 um 01:37 schrieb [EMAIL PROTECTED]:


Author: rgheck
Date: Tue Feb  5 01:37:30 2008
New Revision: 22779

URL: http://www.lyx.org/trac/changeset/22779
Log:
Additional fixes related to disentangling the blinking cursor signal 
from other signals. Thanks to Abdel for pointing out the need for this.


Modified:
   lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp
   lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h

Modified: lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp
URL: 
http://www.lyx.org/trac/file/lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp?rev=22779 

== 


--- lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp (original)
+++ lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp Tue Feb  5 
01:37:30 2008

@@ -211,7 +211,7 @@
general_timer_.setInterval(500);
connect(&general_timer_, SIGNAL(timeout()),
this, SLOT(handleRegularEvents()));
-general_timer_.start();
+startGeneralTimer();

screen_ = QPixmap(viewport()->width(), viewport()->height());
cursor_ = new frontend::CursorWidget();
@@ -333,11 +333,16 @@
void GuiWorkArea::processKeySym(KeySymbol const & key, KeyModifier mod)
{
// In order to avoid bad surprise in the middle of an operation,
-// we better stop the blinking cursor.
+// we better stop the blinking cursor...
stopBlinkingCursor();
+// ...and the general timer.
+stopGeneralTimer();

theLyXFunc().setLyXView(lyx_view_);
theLyXFunc().processKeySym(key, mod);
+   
+// the cursor gets restarted in GuiView::restartCursor()

+startGeneralTimer();
}


@@ -364,11 +369,15 @@
else
cmd = cmd0;

+bool const notJustMovingTheMouse =
+cmd.action != LFUN_MOUSE_MOTION || cmd.button() != 
mouse_button::none;
+   
// In order to avoid bad surprise in the middle of an operation, 
we better stop

-// the blinking cursor.
-if (!(cmd.action == LFUN_MOUSE_MOTION
-&& cmd.button() == mouse_button::none))
+// the blinking cursor and the general timer
+if (notJustMovingTheMouse) {
stopBlinkingCursor();
+stopGeneralTimer();
+}

buffer_view_->mouseEventDispatch(cmd);

@@ -379,15 +388,15 @@
}

// GUI tweaks except with mouse motion with no button pressed.
-if (!(cmd.action == LFUN_MOUSE_MOTION
-&& cmd.button() == mouse_button::none)) {
+if (notJustMovingTheMouse) {
// Slight hack: this is only called currently when we
// clicked somewhere, so we force through the display
// of the new status here.
lyx_view_->clearMessage();

-// Show the cursor immediately after any operation.
+// Show the cursor immediately after any operation
startBlinkingCursor();
+startGeneralTimer();
}
}


Modified: lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h
URL: 
http://www.lyx.org/trac/file/lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h?rev=22779 

== 


--- lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h (original)
+++ lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h Tue Feb  5 
01:37:30 2008

@@ -121,6 +121,10 @@
void stopBlinkingCursor();
///
void startBlinkingCursor();
+///
+void startGeneralTimer() { general_timer_.start(); }
+///
+void stopGeneralTimer() { general_timer_.stop(); }
/// Process Key pressed event.
/// This needs to be public because it is accessed externally by 
GuiView.

void processKeySym(KeySymbol const & key, KeyModifier mod);

Re: important info about the floatflt package and the LyX manuals

[Pavel Sanda]

> Uwe Stöhr wrote:
>>
>> Unfortunately using wrapfig like in trunk cannot be backported since it is 
>> a fileformat change.
> Well - this rule doesn't have to be absolute!

this is not just about the rules. being as user used to increace from x to
1.y.x+1 without any worry about _the latex output_ is very important imho. 
i will not study way too much release notes for such an update. but i will
enquire what this 1.6 change mean, perfectly knowing i could face some problems
otherwise.

pavel

Re: r22779 - in /lyx-devel/trunk/src/frontends/qt4: GuiWorkAr...

[Stefan Schimanski]

Now I get a segfault, when closing a buffer via the keyboard shortcut:

#0  0x024c7197 in QObject::startTimer ()
#1  0x024d0381 in QTimer::start ()
#2  0x0048be48 in lyx::frontend::GuiWorkArea::startGeneralTimer  
(this=0x1a25a030) at qt4/GuiWorkArea.h:125
#3  0x00486e41 in lyx::frontend::GuiWorkArea::processKeySym  
(this=0x1a25a030, [EMAIL PROTECTED], mod=lyx::ControlModifier) at / 
Users/sts/Quellen/mac/lyx-scope-dynmacro/src/frontends/qt4/ 
GuiWorkArea.cpp:345
#4  0x00489376 in lyx::frontend::GuiWorkArea::keyPressEvent  
(this=0x1a25a030, ev=0xbfffe970) at /Users/sts/Quellen/mac/lyx-scope- 
dynmacro/src/frontends/qt4/GuiWorkArea.cpp:737

#5  0x01ac1a48 in QWidget::event ()
#6  0x01d70eb1 in QFrame::event ()
#7  0x01df42f6 in QAbstractScrollArea::event ()
#8  0x00489568 in lyx::frontend::GuiWorkArea::event  
(this=0x1a25a030, e=0xbfffe970) at /Users/sts/Quellen/mac/lyx-scope- 
dynmacro/src/frontends/qt4/GuiWorkArea.cpp:544

@@ -333,11 +333,16 @@
void GuiWorkArea::processKeySym(KeySymbol const & key, KeyModifier  
mod)

{
// In order to avoid bad surprise in the middle of an operation,
-   // we better stop the blinking cursor.
+   // we better stop the blinking cursor...
stopBlinkingCursor();
+   // ...and the general timer.
+   stopGeneralTimer();

theLyXFunc().setLyXView(lyx_view_);
theLyXFunc().processKeySym(key, mod);
+   
+   // the cursor gets restarted in GuiView::restartCursor()
+   startGeneralTimer();
}



This very last call is a bad idea. LFUN can kill the workarea when you  
close the buffer via the keyboard. So we are inside of a method of an  
deallocated object.


Stefan

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

Abdelrazak Younes <[EMAIL PROTECTED]> writes:

> All our investigations converged to the conclusion that -m3 is way too
> much considering our physical RAM. So:
>
> Option 1: let's just try -m2 for now and see if things improve.
>
> Option 2 (preferred): go straight to -m1, wait a few days and see if
> this has resulted in more spam and/or delayed mail delivery. Switch to
> option 1 if this is the case.

Let's see first what happens now that I have restarted. It may be that
163M for one of the process was some kind of accident. Are we sure
that the RSS of these 3 processes are not shared? How do we get to
know that?

JMarc

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

Pavel Sanda <[EMAIL PROTECTED]> writes:

> no, this belongs to spamassasin file.
> we have 
> SPAMDOPTIONS="-d -c -m3 -H"
> so we should change m3 to ... hmm m1?

This is the big question. I have no idea where to look to see how long
the messages take to go through the whole chain.

If I look for the spamd processes, I see:

[EMAIL PROTECTED] ~]$ ps auxw|grep spamd
root  3361  0.0  0.6  29640  1644 ?Ss   Feb01   0:01 /usr/bin/spamd 
-d -c -m3 -H -r /var/run/spamassassin/spamd.pid
root 19132  0.9  8.9  34012 22988 ?S11:48   0:25 spamd child
root 19175  0.7  8.2  32604 21132 ?S11:48   0:20 spamd child
root 19181  1.6 42.4 163516 108572 ?   S11:48   0:45 spamd child

So there are 3 children, but one of them is 5 times larger than the
others. After a restart, I get:
[EMAIL PROTECTED] ~]$ ps auxw|grep spamd
root 20662  0.5  9.8  29640 25244 ?Ss   12:35   0:00 /usr/bin/spamd 
-d -c -m3 -H -r /var/run/spamassassin/spamd.pid
root 20670  0.9 10.3  30936 26380 ?S12:35   0:01 spamd child
root 20671  0.5 10.2  30696 26080 ?S12:35   0:00 spamd child
root 20672  0.2 10.1  30436 25840 ?S12:35   0:00 spamd child

Now the 3 spamd children have a reasonable rss (the column before ?).

Let's wait a bit and see whether this increases again.

JMarc

Re: r22779 - in /lyx-devel/trunk/src/frontends/qt4: GuiWorkAr...

[Stefan Schimanski]

Now I get a segfault, when closing a buffer via the keyboard shortcut:

#0  0x024c7197 in QObject::startTimer ()
#1  0x024d0381 in QTimer::start ()
#2  0x0048be48 in lyx::frontend::GuiWorkArea::startGeneralTimer  
(this=0x1a25a030) at qt4/GuiWorkArea.h:125
#3  0x00486e41 in lyx::frontend::GuiWorkArea::processKeySym  
(this=0x1a25a030, [EMAIL PROTECTED], mod=lyx::ControlModifier) at /Users/ 
sts/Quellen/mac/lyx-scope-dynmacro/src/frontends/qt4/GuiWorkArea.cpp:345
#4  0x00489376 in lyx::frontend::GuiWorkArea::keyPressEvent  
(this=0x1a25a030, ev=0xbfffe970) at /Users/sts/Quellen/mac/lyx-scope- 
dynmacro/src/frontends/qt4/GuiWorkArea.cpp:737

#5  0x01ac1a48 in QWidget::event ()
#6  0x01d70eb1 in QFrame::event ()
#7  0x01df42f6 in QAbstractScrollArea::event ()
#8  0x00489568 in lyx::frontend::GuiWorkArea::event (this=0x1a25a030,  
e=0xbfffe970) at /Users/sts/Quellen/mac/lyx-scope-dynmacro/src/ 
frontends/qt4/GuiWorkArea.cpp:544


Stefan

Am 05.02.2008 um 01:37 schrieb [EMAIL PROTECTED]:


Author: rgheck
Date: Tue Feb  5 01:37:30 2008
New Revision: 22779

URL: http://www.lyx.org/trac/changeset/22779
Log:
Additional fixes related to disentangling the blinking cursor signal  
from other signals. Thanks to Abdel for pointing out the need for  
this.


Modified:
   lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp
   lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h

Modified: lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp
URL: 
http://www.lyx.org/trac/file/lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp?rev=22779
= 
= 
= 
= 
= 
= 
= 
= 
==

--- lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp (original)
+++ lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.cpp Tue Feb  5  
01:37:30 2008

@@ -211,7 +211,7 @@
general_timer_.setInterval(500);
connect(&general_timer_, SIGNAL(timeout()),
this, SLOT(handleRegularEvents()));
-   general_timer_.start();
+   startGeneralTimer();

screen_ = QPixmap(viewport()->width(), viewport()->height());
cursor_ = new frontend::CursorWidget();
@@ -333,11 +333,16 @@
void GuiWorkArea::processKeySym(KeySymbol const & key, KeyModifier  
mod)

{
// In order to avoid bad surprise in the middle of an operation,
-   // we better stop the blinking cursor.
+   // we better stop the blinking cursor...
stopBlinkingCursor();
+   // ...and the general timer.
+   stopGeneralTimer();

theLyXFunc().setLyXView(lyx_view_);
theLyXFunc().processKeySym(key, mod);
+   
+   // the cursor gets restarted in GuiView::restartCursor()
+   startGeneralTimer();
}


@@ -364,11 +369,15 @@
else
cmd = cmd0;

+   bool const notJustMovingTheMouse =
+		cmd.action != LFUN_MOUSE_MOTION || cmd.button() !=  
mouse_button::none;

+   
	// In order to avoid bad surprise in the middle of an operation, we  
better stop

-   // the blinking cursor.
-   if (!(cmd.action == LFUN_MOUSE_MOTION
-   && cmd.button() == mouse_button::none))
+   // the blinking cursor and the general timer
+   if (notJustMovingTheMouse) {
stopBlinkingCursor();
+   stopGeneralTimer();
+   }

buffer_view_->mouseEventDispatch(cmd);

@@ -379,15 +388,15 @@
}

// GUI tweaks except with mouse motion with no button pressed.
-   if (!(cmd.action == LFUN_MOUSE_MOTION
-   && cmd.button() == mouse_button::none)) {
+   if (notJustMovingTheMouse) {
// Slight hack: this is only called currently when we
// clicked somewhere, so we force through the display
// of the new status here.
lyx_view_->clearMessage();

-   // Show the cursor immediately after any operation.
+   // Show the cursor immediately after any operation
startBlinkingCursor();
+   startGeneralTimer();
}
}


Modified: lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h
URL: 
http://www.lyx.org/trac/file/lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h?rev=22779
= 
= 
= 
= 
= 
= 
= 
= 
==

--- lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h (original)
+++ lyx-devel/trunk/src/frontends/qt4/GuiWorkArea.h Tue Feb  5  
01:37:30 2008

@@ -121,6 +121,10 @@
void stopBlinkingCursor();
///
void startBlinkingCursor();
+   ///
+   void startGeneralTimer() { general_timer_.start(); }
+   ///
+   void stopGeneralTimer() { general_timer_.stop(); }
/// Process Key pressed event.
	/// This needs to be public because it is accessed externally by  
GuiView.

void processKeySym(KeySymbol const & key, KeyModifier mod);

Re: Aussie is *slow*

[Pavel Sanda]

> > But first we should make sure that -m2 or -m1 is a good value.
> 
> how do you want to assure in an other way than trying it ? :) 
> i proposed 1 because of seeing one can take 109mb in resident memory
> (and 163516mb in total with swap), so even 2 are enough to swap

163 mb of course :)
p

Re: Aussie is *slow*

[Pavel Sanda]

> But first we should make sure that -m2 or -m1 is a good value.

how do you want to assure in an other way than trying it ? :) 
i proposed 1 because of seeing one can take 109mb in resident memory
(and 163516mb in total with swap), so even 2 are enough to swap
aussie to death.

pavel

Re: important info about the floatflt package and the LyX manuals

[Helge Hafting]

Uwe Stöhr wrote:


Unfortunately using wrapfig like in trunk cannot be backported since 
it is a fileformat change.

Well - this rule doesn't have to be absolute!

It sure is a good rule to have for changes initiated by the LyX project.
But this time the change is imposed from the outside.

Distributing a LyX known not to work is kind of silly - if the only
reason is that we don't want to do a fileformat change. (Nobody
wanting to add wrapfig to 1.5.4 is reasonable though.)

Doing a big job on the installer for a fix that will be obsoleted by
1.6 doesn't seem smart in this case either. No problem if
someone wants to do it anyway - but changing the fileformat to cope
with changes outside LyX is actually a good idea. It is
just the sort of thing maintenance releases are for - sync
up with TeX changes.

Of course there is also the option of renaming 1.5.4 to 1.6 because
of a single file format change, and then 1.6svn becomes 1.7svn.  The
rigid rule is then followed, we get a LyX that works with the latest Tex,
and also a shorter interval between 1.6 and 1.7 :-)

Helge Hafting

Re: Aussie is *slow*

[Abdelrazak Younes]

Jean-Marc Lasgouttes wrote:

[EMAIL PROTECTED] writes:


I'm _guessing_ this is the file to change:
/etc/mail/spamassassin/local.cf

but I don't know what to add...


No, it is /etc/sysconfig/spamassassin:

---
# Options to spamd
SPAMDOPTIONS="-d -c -m3 -H"


But first we should make sure that -m2 or -m1 is a good value.


All our investigations converged to the conclusion that -m3 is way too 
much considering our physical RAM. So:


Option 1: let's just try -m2 for now and see if things improve.

Option 2 (preferred): go straight to -m1, wait a few days and see if 
this has resulted in more spam and/or delayed mail delivery. Switch to 
option 1 if this is the case.


Abdel.

Re: Aussie is *slow*

[Helge Hafting]

Pavel Sanda wrote:
I'm NOT running mailman, spam assassin, or vsftpd, but could those be 
eating 100MB? That seems unlikely. 



spam assassin can be very expensive, 100mb are imho no problem to eat with it.

  

Would moving some services elsewhere help? E.g., moving the ftp server?



i would look on httpd and spam services first.
  

These services _can_ eat lots of memory (and sometimes cpu), and therefore
they also have config options for not doing that.
Log in, see what proess is eating most memory.  Then configure
it to use less - fewer simultaneous processes, less caching/memory
allocation if there are configs for that.  If not - consider different
server sw with more micromanagement options.

If something uses too much cpu - configure for fewer simultaneous
processes and/or use 'nice'.

A service getting less resources may get worse performance,
but everything else will benefit. Swap avoidance may even help
everything: The services getting less resources will now
have a slower "best case response" but perhaps they
never saw that "best case" anyway. Massive improvement of the
worst case can help a lot. :-)

Helge Hafting

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

[EMAIL PROTECTED] writes:

> I'm _guessing_ this is the file to change:
>   /etc/mail/spamassassin/local.cf
>
> but I don't know what to add...

No, it is /etc/sysconfig/spamassassin:

---
# Options to spamd
SPAMDOPTIONS="-d -c -m3 -H"


But first we should make sure that -m2 or -m1 is a good value.

JMarc

Re: important info about the floatflt package and the LyX manuals

[Jean-Marc Lasgouttes]

Uwe Stöhr <[EMAIL PROTECTED]> writes:

>> Is there really a lyx2lyx entry related to this change?
>
> I checked my commit history. When we only use a very basic
> implementation, I think we could backport it:
> http://www.lyx.org/trac/changeset/20185

Yes.

> This was the first commit that changes wrap floats from floatflt to
> wrapfig. It only need a dummy fileformat change since the content of
> the LyX-files is not really changed, only the functionality in the
> background.

I think even the dummy file format change was not needed, since there
is nothing lyx2lyx can do to alleviate whatever problem may appear.
This change of package may change the output, there not much we can do
about it.

I propose to backport a subset of this (even the GUI changes may not
be necessary).

Juergen, as always it s up to you.

JMarc

Re: Aussie is *slow*

[Pavel Sanda]

> I also noticed that we should restart spamd after changing the 
> configuration file, using SIGHUP seems somewhat more tricky.
>
> I'm _guessing_ this is the file to change:
>   /etc/mail/spamassassin/local.cf
>
> but I don't know what to add...

no, this belongs to spamassasin file.
we have 
SPAMDOPTIONS="-d -c -m3 -H"
so we should change m3 to ... hmm m1?

also sighup is not clean way - dont know this particular distro, but 
i usually force services change through some /etc/init.d/service restart
or such.

pavel

Re: Aussie is *slow*

[christian . ridderstrom]

On Tue, 5 Feb 2008, Pavel Sanda wrote:


On Mon, 4 Feb 2008, rgheck wrote:


 looking at the RES memory usage, the main consumers are these. what if
 we start by running less childerns of spamd?

   59m   2:32.53 spamd   57m   0:52.08 spamd
   37m   1:48.72 spamd


Do we even need 'spamd'?   I mean, does it serve a useful purpose on
aussie?


i guess its purpose is to filter spam from [EMAIL PROTECTED] address and not 
doing
this filter would mean megabytes of disk space on aussie every day because many
people dont forward their mails etc.


We should check on which addresses it is run. I suspect there is one
instance per mail address...


afaik spamd has fixed number of childs (here 3); so its just to change this
parameter in config.


Makes sense... I found this for an option that can be given to spamd:

-m number , --max-children=number

This option specifies the maximum number of children to spawn.

minimum value is 1, the default value is 5.


Note that if you run too many servers for the amount of free RAM
available, you run the danger of hurting performance by causing a
high swap load as server processes are swapped in and out
continually.

I also noticed that we should restart spamd after changing the 
configuration file, using SIGHUP seems somewhat more tricky.


I'm _guessing_ this is the file to change:
/etc/mail/spamassassin/local.cf

but I don't know what to add...

/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[Pavel Sanda]

>>>  E1) Putting the section about wrapping floats in a branch?
>>>  (Will this help?)
>>
>> This is indeed an option.
>
> How about the following solution:
>
>   (Temporarily) move the section regarding wrapping floats from the users'
>   guide to some other document, and let the section in the users' guide
>   refer to that document.

reread the thread ;) this is not a solution as not only documentation is 
inflicted.
pavel

Re: kubuntu710 and lyx

[Abdelrazak Younes]

Pavel Sanda wrote:
Well, it's OK (and recommended unless you're a developer) to disable 
stdlib-debug but please don't get rid of -g, at least for 1.6svn as this 
will allow to report useful backtrace in case of crash. There's no speed 
penalty by doing that IIRC, you just get a bigger executable, that's all. 


i) ages to link it ii) ages to run. .. but ok i dont have to much RAM here.
but its true backtraces are useful for reporting problems.


There's a more minimal debug option IIRC, was that -gfabs or something?

Abdel.

Re: kubuntu710 and lyx

[Pavel Sanda]

> Well, it's OK (and recommended unless you're a developer) to disable 
> stdlib-debug but please don't get rid of -g, at least for 1.6svn as this 
> will allow to report useful backtrace in case of crash. There's no speed 
> penalty by doing that IIRC, you just get a bigger executable, that's all. 

i) ages to link it ii) ages to run. .. but ok i dont have to much RAM here.
but its true backtraces are useful for reporting problems.
pavel

Re: Aussie is *slow*

[Pavel Sanda]

> > On Mon, 4 Feb 2008, rgheck wrote:
> >
> >>>  looking at the RES memory usage, the main consumers are these. what if
> >>>  we start by running less childerns of spamd?
> >>>
> >>>59m   2:32.53 spamd   57m   0:52.08 spamd
> >>>37m   1:48.72 spamd
> >
> > Do we even need 'spamd'?   I mean, does it serve a useful purpose on
> > aussie?

i guess its purpose is to filter spam from [EMAIL PROTECTED] address and not 
doing
this filter would mean megabytes of disk space on aussie every day because many
people dont forward their mails etc.

> We should check on which addresses it is run. I suspect there is one
> instance per mail address...

afaik spamd has fixed number of childs (here 3); so its just to change this
parameter in config.

pavel

Re: Aussie is *slow*

[Jean-Marc Lasgouttes]

[EMAIL PROTECTED] writes:

> On Mon, 4 Feb 2008, rgheck wrote:
>
>>>  looking at the RES memory usage, the main consumers are these. what if
>>>  we start by running less childerns of spamd?
>>>
>>>59m   2:32.53 spamd   57m   0:52.08 spamd
>>>37m   1:48.72 spamd
>
> Do we even need 'spamd'?   I mean, does it serve a useful purpose on
> aussie?

We should check on which addresses it is run. I suspect there is one
instance per mail address...

JMarc

Re: Aussie is *slow*

[christian . ridderstrom]

On Mon, 4 Feb 2008, rgheck wrote:


 looking at the RES memory usage, the main consumers are these. what if
 we start by running less childerns of spamd?

   59m   2:32.53 spamd   57m   0:52.08 spamd
   37m   1:48.72 spamd


Do we even need 'spamd'?   I mean, does it serve a useful purpose on 
aussie?


/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: proposal to upgrade Aussie

[christian . ridderstrom]

On Mon, 4 Feb 2008, rgheck wrote:


 - buy more memory ->
   we need to ask someone at Trolltech to build it into the machine

The discussion elsewhere seems to me to suggest that this is the best 
option. For what we're running, it'd be surprising if an Athlon 2000+ 
wasn't more than adequate. The problem seems to be swapping. In any 
event, it will likely be a lot cheaper to install a bit of memory than 
to get a new server.


Then again, we could also use a bigger /var partition, right?


I have a vague memory of a discussion related to this in Finland, and I 
think we at the time thought more memory (RAM + HDD) would be enough.


Let's hope that Lars' sees this, he's the one with knowledge about it 
IIRC.


So the questions would then be:
* Can we upgrade the memory?
* What kind of RAM does it take?
* What kind of HDD could we add to it?
* Is it practially possible for us to do this? Does anyone have physical
  access?

I guess we need someone with physical access in order to even figure out 
what kind of RAM and HDD we could upgrade it with?. At least I am not able 
to figure out this remotely.


/Christian

--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[christian . ridderstrom]

On Tue, 5 Feb 2008, Uwe Stöhr wrote:


 E1) Putting the section about wrapping floats in a branch?
 (Will this help?)


This is indeed an option.


How about the following solution:

  (Temporarily) move the section regarding wrapping floats from the users'
  guide to some other document, and let the section in the users' guide
  refer to that document.

This document could even be a wiki page, and once 1.6 comes out, we put 
the section back in the users' guide.


This extra document can also explain how to install floatflt etc. If we go 
with this solution, I'll volunteer to set up the wiki page. Actually, I 
could just do it as a self-explaining example... complete with .lyx-file 
and .pdf-file to be downloaded.


/Christian

PS. This is embarrassing, but I heard that the documentation _is_ 
available as PDF... Wheredo you find it?   And perhaps more importantly, 
_how_ do you find it if you didn't know that it existed and thus didn't 
think of searching for it?


Is there some part of the introduction or tutorial that explains that 
these documents are also available as PDFs, and where they can be found?



--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr

Re: important info about the floatflt package and the LyX manuals

[christian . ridderstrom]

On Tue, 5 Feb 2008, Jean-Marc Lasgouttes wrote:


Uwe Stöhr <[EMAIL PROTECTED]> writes:


Jean-Marc Lasgouttes schrieb:

> The problem is not only the documentation, but a feature of LyX that

does not work anymore. Should we investigate backporting the wrapfig
patch to branch?


But this would be a fileformat change.


Sure, but having files suddenly not typesetting is also a file format 
change :( Is there really a lyx2lyx entry related to this change?


Could we do a 1.6.0 release that only has this "backport"?

/Christian


--
Christian Ridderström, +46-8-768 39 44   http://www.md.kth.se/~chr