Re: [Rosegarden-bugs] SF.net SVN: rosegarden:[11212] trunk/rosegarden

[Heikki Johannes Junes]

2009/11/10 <janifr@...>

Revision: 11212
         http://rosegarden.svn.sourceforge.net/rosegarden/?rev=11212&view=rev
Author:   janifr
Date:     2009-11-10 16:20:14 +0000 (Tue, 10 Nov 2009)

Log Message:
-----------
Commented out 2 obsolete slots.

Modified Paths:
--------------
   trunk/rosegarden/data/locale/fi.qm

(clip)

Modified: trunk/rosegarden/data/locale/fi.qm
===================================================================
(Binary files differ)

(clip)

Just to check, should the compiled binary translation files be included to the version control?

--

Heikki

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Jani Frilander]

2009/11/10 Heikki Johannes Junes <hjunes@...>

Just to check, should the compiled binary translation files be included to the version control?

They seem to be. IMO there is no reason though.

-Jani


------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[D. Michael McIntyre]

On Tuesday 10 November 2009, Jani Frilander wrote:

> 2009/11/10 Heikki Johannes Junes <hjunes@...>
>
> > Just to check, should the compiled binary translation files be included
> > to the version control?
> >
> They seem to be. IMO there is no reason though.

Yes, there is a reason.  The .qm files must be included in the resource bundle
for translations to function, and everything included in the resource bundle
must be under Subversion version control.  This policy is in effect to ensure
that all files referred to by data.qrc will exist locally, which prevents the
build from failing due to missing data files.

Prior to the implementation of this policy, there were several instances of
someone committing a modified data.qrc without committing new data files.  If
data.qrc refers to a file that does not exist, the build fails, and it is
necessary to waste time hacking around this irritation instead of getting work
done.
--
D. Michael McIntyre

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Heikki Johannes Junes]

2009/11/10 D. Michael McIntyre <michael.mcintyre@...>

On Tuesday 10 November 2009, Jani Frilander wrote:

> 2009/11/10 Heikki Johannes Junes <hjunes@...>
>
> > Just to check, should the compiled binary translation files be included
> > to the version control?
> >
> They seem to be. IMO there is no reason though.

Yes, there is a reason.  The .qm files must be included in the resource bundle
for translations to function, and everything included in the resource bundle
must be under Subversion version control.  This policy is in effect to ensure
that all files referred to by data.qrc will exist locally, which prevents the
build from failing due to missing data files.

Prior to the implementation of this policy, there were several instances of
someone committing a modified data.qrc without committing new data files.  If
data.qrc refers to a file that does not exist, the build fails, and it is
necessary to waste time hacking around this irritation instead of getting work
done.

I have thought that compiled files would not be included to version control,
if source files already exists there. I am reffering to the concept of
"database normalization" [1].

[1] http://en.wikipedia.org/wiki/Database_normalization

Of course, it is better to get work done than stubbornly stick to design patterns.

--

Heikki

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Jani Frilander]

2009/11/10 D. Michael McIntyre <michael.mcintyre@...>

Yes, there is a reason.  The .qm files must be included in the resource bundle
for translations to function, and everything included in the resource bundle
must be under Subversion version control.  This policy is in effect to ensure
that all files referred to by data.qrc will exist locally, which prevents the
build from failing due to missing data files.

I probably don't understand this completely, but doesn't our build system
always recompile the .qm file if the corresponding .ts has changed?

A little test:

frilande@masiina:~/Ohjelmointi/rosegarden$ touch data/locale/fi.ts
frilande@masiina:~/Ohjelmointi/rosegarden$ make
*** WARNING: One or more of Makefile.in, configure, configure.ac and/or
             acinclude.m4 has been modified since the Makefile was generated;
             consider running ./configure again
/usr/bin/lrelease-qt4 data/locale/fi.ts
Updating 'data/locale/fi.qm'...
    Generated 3991 translation(s) (3990 finished and 1 unfinished)

    Ignored 64 untranslated source text(s)
/usr/bin/rcc -threshold 0 -compress 9 data/data.qrc -o data/data.cpp

It seems to rely on timestamps. I need to rerun ./configure...

-Jani

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[D. Michael McIntyre]

On Wednesday 11 November 2009, Jani Frilander wrote:

> I probably don't understand this completely, but doesn't our build system
> always recompile the .qm file if the corresponding .ts has changed?

Yes, but we no longer install anything except MIME-related files.  Everything
else is included in the resource bundle, and unpacked from there at runtime.

In order for the translations to function, the .qm files need to be available
at runtime, and in order to make them available, they have to be compiled into
the resource bundle.

In order for them to be compiled into the resource bundle, data.qrc must refer
to them, and if data.qrc refers to them, they are required to be under version
control as a matter of policy.  Therefore the .qm files need to be under
version control.

The underlying reason for this policy is because someone committed a data.qrc
that referred to a number of files he forgot to add, and then he disappeared
for three days.  This policy is a guarantee against that ever happening again.

The only other alternative I can think of is to not keep data.qrc itself under
version control.  That is possible, but I'm reluctant to go that route,
because we can't ensure consistency from developer to developer if every local
copy builds with whatever random data files happen to be in that developer's
tree.  On my own machine, this would include all kinds of random development-
related files that are intentionally *not* under version control, like icon
templates and things of that general nature.  That would potentially make my
Rosegarden behave differently from your Rosegarden.  Not a big deal if you are
aware of this and you're thinking about it, but then six months later you're
trying to solve some strange icon size problem and then after wasting six
hours you realize it's because you have a local file that isn't part of the
repository, and that's why something is working for you but everyone else has
a strange problem.

I think keeping the .qm files under version control is an acceptable price to
pay for this kind of insurance against stupid time wasting consistency
problems.
--
D. Michael McIntyre

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Jani Frilander]

2009/11/11 D. Michael McIntyre <michael.mcintyre@...>

Yes, but we no longer install anything except MIME-related files.  Everything
else is included in the resource bundle, and unpacked from there at runtime.

 

I think keeping the .qm files under version control is an acceptable price to
pay for this kind of insurance against stupid time wasting consistency
problems.

Thanks for your explanation. Now I think I understand it. Now I remember there
was discussion about this a few months ago on this list.

-Jani

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Heikki Johannes Junes]

2009/11/11 Jani Frilander <j.frilander@...>

2009/11/11 D. Michael McIntyre <michael.mcintyre@...>

I think keeping the .qm files under version control is an acceptable price to
pay for this kind of insurance against stupid time wasting consistency
problems.

Thanks for your explanation. Now I think I understand it. Now I remember there
was discussion about this a few months ago on this list.

I strongly suggest to maintain these kind of consistency checks of the source tree
in the source tree.

I was just thinking of a less surprisingly way of making the consistency checks.
By less suprisingly I mean that the source tree would not contain twice exactly
the same information. If the same information needs to be updated in two places
in the source tree, that leads to an another kind of consistency or dependence flaws.

I have bad bad personal experience on updating some information in one place and
forgetting to update the same information in the second place. It resembles very much
the situation which Michael described. Those kind of double-presence errors can be
sometimes tricky to solve out, because after looking at the first place you may obtain
an impression that everything is ok, but the second place is leaking.

Would anybody hurt if I started to collect a list of "resign patterns" which the project
has adopted? By "resign patterns" I mean coding solutions which have been proven out
to work, but which may have some weak points in the long run. Such preparation could
give a kind of structural approach for improving the internal design of the program.

I think for Thorn we should not change the design too much, but we could prepare for
some design changes for the releases after Thorn.

--

Heikki

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[D. Michael McIntyre]

On Wednesday 11 November 2009, Heikki Johannes Junes wrote:

> I strongly suggest to maintain these kind of consistency checks of the
> source tree in the source tree.

So noted.
 
> Would anybody hurt if I started to collect a list of "resign patterns"
> which the project has adopted?

You can make all the lists you want, and we can take a good look at what you
think needs to be changed, but we have very nearly already committed suicide
as a project already, and the most important priority we have right now is to
SHIP CODE!

Also, be aware that I'm going to announce a feature freeze very soon.
--
D. Michael McIntyre

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[Heikki Johannes Junes]

2009/11/12 D. Michael McIntyre <michael.mcintyre@...>

On Wednesday 11 November 2009, Heikki Johannes Junes wrote:

> Would anybody hurt if I started to collect a list of "resign patterns"
> which the project has adopted?

You can make all the lists you want, and we can take a good look at what you
think needs to be changed,

The motherboard of my Linux machine died and now I finally have only Windows only machines. Hopefully I can do some specification work.
 

but we have very nearly already committed suicide
as a project already,

The story resembles the mythology of the phoenix firebird :)

http://en.wikipedia.org/wiki/Phoenix_%28mythology%29
 

and the most important priority we have right now is to
SHIP CODE!

Also, be aware that I'm going to announce a feature freeze very soon.

Great progress! I was also also suggesting not to change the code base too much. But you are the one who blows the whistle in order to everybode "Freeze!".

Heikki

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel

[D. Michael McIntyre]

On Thursday 12 November 2009, Heikki Johannes Junes wrote:

> The motherboard of my Linux machine died and now I finally have only
>  Windows only machines. Hopefully I can do some specification work.

Work on figuring out how to make Rosegarden run on Windows too.

Although after what Chris said about Audacity, it makes me think twice about
whether I really want Rosegarden on Windows or not.  Big support nightmare.
 
> The story resembles the mythology of the phoenix firebird :)

Indeed, I've been half seriously thinking of this as "Rosegarden Phoenix" for
a few months now.  This has taken more time than our most pessimistic
estimates.

> Great progress! I was also also suggesting not to change the code base too
> much. But you are the one who blows the whistle in order to everybode
> "Freeze!".

Not yet, but it's coming soon.  I've been grinding that way for some untold
number of hours now, and I'm almost there.  As soon as I finish the Spanish
translation and the string audit I'm doing along the way, I'm going to
announce a string freeze, and start thinking very seriously about getting an
alpha tarball release out there.

This is what we need to focus on until it's done and away.  This, and
documentation.
--
D. Michael McIntyre

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Rosegarden-devel mailing list
Rosegarden-devel@... - use the link below to unsubscribe
https://lists.sourceforge.net/lists/listinfo/rosegarden-devel