Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-02 Thread tchize
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

I do not agree on this change, those informations are not so verbose
and have no reason to make users afraid of as they are not visible by
default. Normal users don't start game in console and those starting
in console want basic messages about what is happening. Moreover, this
is pretty usefull to ask on irc a user to dump us content of help -
bugreport when trying to support him. It's easier to ask player to go
to a menu entry then ask him to go in consle mode, and it works
crossplatform

Crossfire CVS repository messages. a écrit :

> Module Name: client Committed By: mwedel Date: Sun Jul 2 03:19:43
> UTC 2006
>
> Modified Files: client: ChangeLog client/common: misc.c
>
> Log Message: common/misc.c: Make default log level 2 when not in
> debug mode. Normal users probably don't want all the INFO log
> messages, and it never makes a good impression about
> stability/quality if a program spews out lots of errors or other
> messages. MSW 2006-07-01
>
>
>

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFEp7ZlHHGOa1Q2wXwRAsHYAJwNpQ3aGDts/IfYbJdPYeGvX6VFEACeJoky
wxfJo+dw4uAaI3ySSS6aYYI=
=fnnG
-END PGP SIGNATURE-


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-02 Thread Mark Wedel
tchize wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
> 
> I do not agree on this change, those informations are not so verbose
> and have no reason to make users afraid of as they are not visible by
> default. Normal users don't start game in console and those starting
> in console want basic messages about what is happening. Moreover, this
> is pretty usefull to ask on irc a user to dump us content of help -
> bugreport when trying to support him. It's easier to ask player to go
> to a menu entry then ask him to go in consle mode, and it works
> crossplatform

  Given (IIRC) that the client does not currently register itself with gnome of 
KDE, I'd expect most people to actually run from the command line.

  Any my experience is that very few programs will dump out all that 
information 
by default.  And one problem in doing so is that it can be harder to see real 
error messages.

  Probably the correct solution is to add a -V or --version option to the 
clients which dump out that information - that is what most all other 
applications do.  It hides the information people don't need by default, but 
still provides a way to get that information in case of bug reports, etc.


> 
> Crossfire CVS repository messages. a écrit :
> 
>> Module Name: client Committed By: mwedel Date: Sun Jul 2 03:19:43
>> UTC 2006
>>
>> Modified Files: client: ChangeLog client/common: misc.c
>>
>> Log Message: common/misc.c: Make default log level 2 when not in
>> debug mode. Normal users probably don't want all the INFO log
>> messages, and it never makes a good impression about
>> stability/quality if a program spews out lots of errors or other
>> messages. MSW 2006-07-01
>>
>>
>>
> 
> -BEGIN PGP SIGNATURE-
> Version: GnuPG v1.4.1 (GNU/Linux)
> 
> iD8DBQFEp7ZlHHGOa1Q2wXwRAsHYAJwNpQ3aGDts/IfYbJdPYeGvX6VFEACeJoky
> wxfJo+dw4uAaI3ySSS6aYYI=
> =fnnG
> -END PGP SIGNATURE-
> 
> 
> ___
> crossfire mailing list
> crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-03 Thread Yann Chachkoff
>   Given (IIRC) that the client does not currently register itself with
> gnome of KDE, I'd expect most people to actually run from the command line.
>
>   Any my experience is that very few programs will dump out all that
> information by default.  And one problem in doing so is that it can be
> harder to see real error messages.
>
It sounds like backward thinking to me. That most people have to run the game 
from the command line (and thus, see all what the client can spit out) seems 
to be the point to solve, and not the messages provided.

Moreover, if the error messages are hard to spot, then make them more obvious 
(add stars around them, write a couple of !!!, shift them, add blank lines 
before and after them, etc). I'd also underline that the error messages 
targetted at the player shouldn't only go to the console, but also be clearly 
visible from the GUI itself in the first place whenever possible. 

In such a case, the player would logically first report the short error 
message he/she saw in a dialog box. Then, if details are needed, he/she can 
sends the whole content of the console messages. Sounds a much better way to 
make at least common errors easier to spot for the player than completely 
hiding what's basically useful information.

>   Probably the correct solution is to add a -V or --version option to the
> clients which dump out that information - that is what most all other
> applications do.  It hides the information people don't need by default,
> but still provides a way to get that information in case of bug reports,
> etc.
>
The question is: which informations are needed by default ? I think the "Info" 
should have an obvious meaning: provide a couple of informations that *may* 
be useful in several cases, debugging being one of them.

If some informations are judged superlfluous at the default level of log 
(typically, what's required only for debugging/post-mortem analysis) then 
those should be put back in the "Debug" log level, which would appear only 
when the verbosity is explicitely increased.

Finally, I don't get why this change was done in the first place - I've so far 
never heard anybody bothered by the amount of log messages provided in the 
console; besides that, there are relatively few messages displayed, even at 
the "Info" level. So not only does the change of the default verbosity level 
in the client sound detrimental to me, but it solves a problem that didn't 
even exist in the first place.

>
> > Crossfire CVS repository messages. a écrit :
> >> Module Name: client Committed By: mwedel Date: Sun Jul 2 03:19:43
> >> UTC 2006
> >>
> >> Modified Files: client: ChangeLog client/common: misc.c
> >>
> >> Log Message: common/misc.c: Make default log level 2 when not in
> >> debug mode. Normal users probably don't want all the INFO log
> >> messages, and it never makes a good impression about
> >> stability/quality if a program spews out lots of errors or other
> >> messages. MSW 2006-07-01
> >
> > -BEGIN PGP SIGNATURE-
> > Version: GnuPG v1.4.1 (GNU/Linux)
> >
> > iD8DBQFEp7ZlHHGOa1Q2wXwRAsHYAJwNpQ3aGDts/IfYbJdPYeGvX6VFEACeJoky
> > wxfJo+dw4uAaI3ySSS6aYYI=
> > =fnnG
> > -END PGP SIGNATURE-
> >
> >
> > ___
> > crossfire mailing list
> > crossfire@metalforge.org
> > http://mailman.metalforge.org/mailman/listinfo/crossfire
>
> ___
> crossfire mailing list
> crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire

-- 
Yann Chachkoff
---
GPG Key 2006: http://keyserver.veridis.com:11371/export?id=-43113965597490782
Fingerprint : C224 F1F9 9025 4FC7 987D 05BB FF66 D413 A3B4 01A2
---
"They misunderestimated me."
 - George W. Bush


pgphImGpG7EyP.pgp
Description: PGP signature
___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-04 Thread Mark Wedel

  General quick thoughts:

  I removed them as I believe (rightly or wrongly) that software presented for 
general end user consumption should not be producing debug or information 
messages - it should only print error messages.  I'd almost be tempted to say 
that the default log level should even go one step higher, so that warnings 
should by default not be printed - only print error or above

  Off the top of my head, I can not think of any other unix/linux programs that 
produce a list of the rcsid versions of all the files when started.  I'm sure 
there are exceptions, but I think it is better of we behave like most other 
programs.

  I'm not even sure if the reporting of rcsid information has ever proved 
useful 
- I can't recall any bug reports where that has been provided.  Most useful bit 
of information that users can provide is the overal version number of the 
client 
they are using, not the version information of the specific files.

  A lot of this is also perception.  While perhaps no official complaints 
registered, I don't think it puts a good perception on the program.

  All that said, thoughts on how to fix this:

1) Within the client itself, this information should be presented in the GUI. 
For the gtk client, I'm going to modify the bug report page to include this 
information.  For the gtk2 client, I'll make an about page that also includes 
that.

  If in fact most people are running the client from a launcher and not seeing 
the messages, then they aren't doing any good (that seems like counter argument 
- no one will see the messages, so having them there doesn't harm anything).  
So 
having an in-client way to display them is better.

2) The log level should not be a compiled in default - one should be able to 
specify it via command line (-loglevel 0) or whatever.  In that case, if user 
has a bug, we can always say 'run with -loglevel 0' and provide us the output. 
This is more useful in all cases, as by default, old log level was 1 so we 
couldn't get debug level output without requesting the user to recompile (which 
doesn't work when it is prebundled).  This should be pretty easy to do I think.

  How do those remedies sound to people?



___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-04 Thread Tchize
Mark Wedel wrote:
>   General quick thoughts:
>
>   I removed them as I believe (rightly or wrongly) that software presented 
> for 
> general end user consumption should not be producing debug or information 
> messages - it should only print error messages.  I'd almost be tempted to say 
> that the default log level should even go one step higher, so that warnings 
> should by default not be printed - only print error or above
>   
 From info to critical is everything that might concern user, every 
thing below is of concern to developpers.
>   Off the top of my head, I can not think of any other unix/linux programs 
> that 
> produce a list of the rcsid versions of all the files when started.  I'm sure 
> there are exceptions, but I think it is better of we behave like most other 
> programs.
>   
This was (probably wrongly) introduced because i couldn't find a better 
way to identify version when it comes to cvs build clients. However, 
these informations can probably go in debug level. At least, the version 
number of client should be increased in cvs just after each release. 
(example: set 1.9.1, commit, tag cvs, release, set to 'post 1.9.1', commit)
>
> 1) Within the client itself, this information should be presented in the GUI. 
> For the gtk client, I'm going to modify the bug report page to include this 
> information.  For the gtk2 client, I'll make an about page that also includes 
> that.
>
>   If in fact most people are running the client from a launcher and not 
> seeing 
> the messages, then they aren't doing any good (that seems like counter 
> argument 
> - no one will see the messages, so having them there doesn't harm anything).  
> So 
> having an in-client way to display them is better.
>   
help -> bug report, there is already everything needed there to provide 
information for a bug report, that is copy of log messages and a link to 
sourceforge tracker...
btw, a 'fork at begin, watch for dying of client process, expose a bug 
report window' might be interresting too, but quite more cumbersome to 
make :)
> 2) The log level should not be a compiled in default 
>   
agree, but info level should be the default.


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-04 Thread Mark Wedel

  One thing which may make sense is different default levels based on if it is 
a 
precompiled binary vs something checked out from CVS.

  I still stand by that the precompiled binaries should be less verbose on 
problems - we're putting this out there as this is quality software, so it 
should act like it.  Perhaps even the official source code releases should also 
be more quiet.

  However, if running from CVS, then having all sorts of error messages is fine.

  I'm not sure if there is any way of automatically doing this detection logic 
in the build environment itself.  I suppose the file can be changed just prior 
to release.  Maybe the best method is also an ability to pass it in via command 
line/CFLAGS (-DLOGLEVEL=...)


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-05 Thread Tchize
Maybe a ./configure --release=
:)
Mark Wedel wrote:
>   One thing which may make sense is different default levels based on if it 
> is a 
> precompiled binary vs something checked out from CVS.
>
>   I still stand by that the precompiled binaries should be less verbose on 
> problems - we're putting this out there as this is quality software, so it 
> should act like it.  Perhaps even the official source code releases should 
> also 
> be more quiet.
>
>   However, if running from CVS, then having all sorts of error messages is 
> fine.
>
>   I'm not sure if there is any way of automatically doing this detection 
> logic 
> in the build environment itself.  I suppose the file can be changed just 
> prior 
> to release.  Maybe the best method is also an ability to pass it in via 
> command 
> line/CFLAGS (-DLOGLEVEL=...)
>
>
> ___
> crossfire mailing list
> crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire
>
>   


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-05 Thread Alex Schultz
Tchize wrote:
> Maybe a ./configure --release=
> :)
I really like that idea, except for one issue. People compiling from
source releases typically wouldn't do that, so I think that what would
be better than including that in the ./configure, would be having a
simple script that could be run as part of the release process.

Alex Schultz

___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-05 Thread Mark Wedel
Alex Schultz wrote:
> Tchize wrote:
>> Maybe a ./configure --release=
>> :)
> I really like that idea, except for one issue. People compiling from
> source releases typically wouldn't do that, so I think that what would
> be better than including that in the ./configure, would be having a
> simple script that could be run as part of the release process.

  Updating the version number requires the update of the configure.ac script 
(then it builds the correct package, etc).  Likewise for the rpmspec file.

  However, doing something like ./configure --debuglevel=0 does work, because 
the rpmspec, which at least is used for rpm building, does have the configure 
line it uses, so modifying that to include the debug option works just fine and 
dandy.

  I'll look at doing that.

___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-05 Thread Wim Villerius
maybe add in the configure script something like (pseudo code)
if -e "./CVS" then
cvs-date=date_of_"./CVS"

that won't give a version number, but it does include a clue about the
version used. 



On Wed, 2006-07-05 at 09:54 -0600, Alex Schultz wrote:
> Tchize wrote:
> > Maybe a ./configure --release=
> > :)
> I really like that idea, except for one issue. People compiling from
> source releases typically wouldn't do that, so I think that what would
> be better than including that in the ./configure, would be having a
> simple script that could be run as part of the release process.
> 
> Alex Schultz
> 
> ___
> crossfire mailing list
> crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire
> 


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-05 Thread Mark Wedel
Mark Wedel wrote:
> Alex Schultz wrote:
>> Tchize wrote:
>>> Maybe a ./configure --release=
>>> :)
>> I really like that idea, except for one issue. People compiling from
>> source releases typically wouldn't do that, so I think that what would
>> be better than including that in the ./configure, would be having a
>> simple script that could be run as part of the release process.
> 
>   Updating the version number requires the update of the configure.ac script 
> (then it builds the correct package, etc).  Likewise for the rpmspec file.
> 
>   However, doing something like ./configure --debuglevel=0 does work, because 
> the rpmspec, which at least is used for rpm building, does have the configure 
> line it uses, so modifying that to include the debug option works just fine 
> and 
> dandy.
> 
>   I'll look at doing that.

  That was easier than expected, so I put in a --with-logevel=x configure 
options.


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client

2006-07-06 Thread Tchize
Let's just do what Wim Villerius suggested:
if we find a .CVS, use it as version number, else use version number 
read from a simple VERSION file.
This all can be calculated at very begining of configure script, before 
defining version number :)
Mark Wedel wrote:
> Alex Schultz wrote:
>   
>> Tchize wrote:
>> 
>>> Maybe a ./configure --release=
>>> :)
>>>   
>> I really like that idea, except for one issue. People compiling from
>> source releases typically wouldn't do that, so I think that what would
>> be better than including that in the ./configure, would be having a
>> simple script that could be run as part of the release process.
>> 
>
>   Updating the version number requires the update of the configure.ac script 
> (then it builds the correct package, etc).  Likewise for the rpmspec file.
>
>   However, doing something like ./configure --debuglevel=0 does work, because 
> the rpmspec, which at least is used for rpm building, does have the configure 
> line it uses, so modifying that to include the debug option works just fine 
> and 
> dandy.
>
>   I'll look at doing that.
>
> ___
> crossfire mailing list
> crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire
>
>   


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client - info level

2006-07-02 Thread tchize
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Most distros provide a shortcut menu to client, moreover, windows
users don't use the console and it would be very difficult to have
them running a program from console to pass log argument. I ran it in
info mode just to check, all i got is 30 lines of info telling user
mainly about client modules version and server communication. This is
not so much, i remember games like half-life or UT dump pages of
messages in console and nobody is afraid of this. And i don't see
where this make errors difficulut to find. By user? When user has
program running like he wants, am pretty sure it doesn't read console!
By developper? developper will need at least an info level to see
what's wrong most of the time.

Seriously, I don't want to fight on this point, this is pretty trivial
and i have othere things to do. This is my last mail on this subjectn
do what you like. I have never seen a player on irc being afraid of
those messages and i have seen several time it proves usefull to
detect version of client / detect kind of server the user is playing
on. Usefull informations are provided on info level while not slowing
client and not invading user. There was no point removing that
behaviour. There is no gain in removal and there is informations lost.

Greetings,
Tchize
Mark Wedel a écrit :

> tchize wrote:
>

> I do not agree on this change, those informations are not so
> verbose and have no reason to make users afraid of as they are not
> visible by default. Normal users don't start game in console and
> those starting in console want basic messages about what is
> happening. Moreover, this is pretty usefull to ask on irc a user to
> dump us content of help - bugreport when trying to support him.
> It's easier to ask player to go to a menu entry then ask him to go
> in consle mode, and it works crossplatform
>
>
>> Given (IIRC) that the client does not currently register itself
> with gnome of
>> KDE, I'd expect most people to actually run from the command
>> line.
>
>> Any my experience is that very few programs will dump out all
> that information
>> by default. And one problem in doing so is that it can be harder
>>
> to see real
>> error messages.
>
>> Probably the correct solution is to add a -V or --version option
> to the
>> clients which dump out that information - that is what most all
>> other applications do. It hides the information people don't
>> need by
> default, but
>> still provides a way to get that information in case of bug
> reports, etc.
>
>
> Crossfire CVS repository messages. a écrit :
>
>> Module Name: client Committed By: mwedel Date: Sun Jul 2 03:19:43
>> UTC 2006
>
>> Modified Files: client: ChangeLog client/common: misc.c
>
>> Log Message: common/misc.c: Make default log level 2 when not in
>> debug mode. Normal users probably don't want all the INFO log
>> messages, and it never makes a good impression about
>> stability/quality if a program spews out lots of errors or other
>> messages. MSW 2006-07-01
>
>
>

___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire



> ___ crossfire mailing
> list crossfire@metalforge.org
> http://mailman.metalforge.org/mailman/listinfo/crossfire



-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFEqDjdHHGOa1Q2wXwRAqCsAKC98YV4k/5tWZMbolb2E7egpZGkmACeJDh1
838DvvJQbVxXYeMf07BS13M=
=ZSSC
-END PGP SIGNATURE-


___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire


Re: [crossfire] [Crossfire-cvs] CVS commit: client - info level

2006-07-03 Thread Nicolas Weeger (Laposte)
Hello.

> Most distros provide a shortcut menu to client, moreover, windows
> users don't use the console and it would be very difficult to have
> them running a program from console to pass log argument. I ran it in

Actually, Windows client does get console. Shortcut runs app, and you see the 
console still.

Nicolas

___
crossfire mailing list
crossfire@metalforge.org
http://mailman.metalforge.org/mailman/listinfo/crossfire