Testing Sylpheed 2.2.4 for Puppy 1.09 (Done)

[pakt]

I would like to test Sylpheed 2.2.4 for Puppy 1.09 to see if a bug with i18n (

[MU]

I can't compile it, get an error:
can't compute sizeof (unsigned shortint)

So I tried the debian-package.

It needed these, + several ones from my Gnome-mini:
libaspell15_0.60.2+20050121-2_i386.deb
libcompfaceg1_1989.11.11-24_i386.deb
libgpgme11_1.0.2-1_i386.deb
libgtkspell0_2.0.10-1_i386.deb
libldap2_2.1.30-8_i386.deb
libpisock8_0.11.8-10_i386.deb
libreadline4-dbg_4.3-11_i386.deb
libreadline5-dbg_5.0-10_i386.deb
libsasl2_2.1.19-1.5sarge1_i386.deb
libssl0.9.8_0.9.8a-8_i386.deb

Extracted they are 13 MB (without gnome-mini).

So what I could do:
rebuild gnome-mini with sylpheed and the libs.
This packages will need another update in future, with all the updated developper-libs, or you could compile no more Gtk-apps.

Tell me, if that interests you.
You also could send me an email with the chars not displayed correctly, so I can look, if it is ok.

gnome-mini:
http://www.murga.org/~puppy/viewtopic.php?t=7671

Mark

[MU]

ok, here it is

http://www.murga.org/~puppy/viewtopic.php?p=47598#47598

Mark

[pakt]

Thanks Mark for going through the trouble

Hmm, seems like good news and bad news...

The good news: The problem with i18n characters has been fixed in Sylpheed 2.2.4. So, speaking for myself and others that may have had the i18n problem, this version would be fine in Puppy 1.09.

The bad news: If I understand correctly, Sylpheed 2.2.4 cannot simply replace 2.0.1 in Puppy 1.09. Because of Gnome dependencies, the size of Sylpheed would increase to at least 13MB (extracted). Without having the figures at hand, wouldn't that make Sylpheed about the size of Thunderbird ??

Paul

[MU]

no, sylpheed 2.2.4 could replace the old one, if someone could find out, how to avoid the error I had when compiling it (shortint).
Maybe I just messed up my system with all the stuff I tested in the past.

As I could not compile, I had to use a already compiled binary.
In Debian, it was just available from Debian testing, that is compiled with a new version of Gtk, that uses new libraries like Cairo.
Cairo allows to use scalable widgets, an almost unused feature.
But once a binary is compiled with these libs, you need them.
It was also compiled with a new glib, that is a central part of every system.
Updating it, requires to update many other libs, too.

I had this problem too, with gnome. In fact, I initially just wanted a small program called Gtkhtml, that did not compile in Puppy. After I found out, what libraries I need for the Debian-binary, it was just a "small" addon, to add some more to get gnome working.
For this reason I stuffed all these new libs together in a "Gnome"-addon.

So conclusion:
You might try to compile it yourself, or try to find a binary compiled on other systems, maybe slackware or fedora3.
That might work with Standard-puppy, I was just too lazy to search the different repositories myself

Mark

[pakt]

If someone like you with years of experience with C can't compile Sylpheed, what makes you think that someone like me with almost zero C experience can succeed ?

Seriously, I have tried to learn C but with my health problems (chronic fatigue), I'm too tired most of the time and have trouble concentrating. I used to do some programming in the days when Pascal was _the_ language. I liked it because compiling a program either succeeded or produced some error messages that helped debug the program.

With C, I got put off early on by nearly always getting a bunch of cryptic compilation errors even when compiling succeeded -- that didn't do my self-confidence much good

For now I'll just stick with being a Puppy tester.

[rarsa]

What makes you think that someone like me with almost zero C experience can succeed ?

Good news for people affraid of compiling.

You don't need to know how to program in any language (let alone in C) to compile a program.

Actually the position of 'source configuration management' in many companies is staffed by people that haven't written a single line of code in the application they prepare for build.

Cheers

[pakt]

Good news for people affraid of compiling.

You don't need to know how to program in any language (let alone in C) to compile a program.

Thanks for the words of encouragement rarsa, but what happens when/if the compilation goes wrong. Having followed many forum threads, this doesn't seem to be all that rare when it comes to C. Just look at MU's post above mine -- what would a non-programmer make of/do with the error message he got, eh ?

Paul

[MU]

I just got the compilation-errors, as by default it wanted to compile with some nonexisting libraries.
With the right options, it worked

No more need for the Gnome-version:
http://www.murga.org/~puppy/viewtopic.php?p=47758#47758

Mark

[pakt]

I just got the compilation-errors, as by default it wanted to compile with some nonexisting libraries.
With the right options, it worked

You just confirmed what I was trying to say: It takes a programmer to sort out compilation errors.

You have just helped improve Puppy 1.09 as a developer by compiling a version of Sylpheed without the bug I found as a tester.

Nice work

[rarsa]

You just confirmed what I was trying to say: It takes a programmer to sort out compilation errors.

No, he just confirmed that you don't need to look at the code. You don't need to know a single thing about C or programming.

The messages may look cryptic not because you are compiling C, but because you haven't seen them before.

But any way, maybe the missunderstanding is in 'what is a programmer'. For me the person that writes the program is the programmer. The compilation is prepared by the 'configuration manager' and the person that executes the compilation in his/her own computer is the user (or tester). Sometimes they are the same, but sometimes the programer does not know al the intricacies of the make process and the 'configuration manager' hasn't even seen the code.

Finally I completelly agree with you: user's (testers) shouldn't need to know to either program or understand the make process.

[pakt]

Interesting to hear of a new catagory: 'configuration manager'. The programmers I worked with did all the work themselves -- from writing the code to compiling and distributing it. Of course, the company I worked for manufactured hardware as their main product. Software came second.

user's (testers) shouldn't need to know to either program or understand the make process.

Perhaps there is a missunderstanding here. Testers are the quality assurance people that systematically test a product (software in this case) to see that it works as it should before it is released. This is (or should be) part of the development cycle.

Users on the other hand are the customers of the software. They get it when it has been released. While it is true that users find bugs, this only happens in the course of actually using the software.

You equate users with testers, which I find interesting. A good friend of mine is a quality assurance boss at the international company I used to work for. She often expresses frustration over the fact that her department has so low regard. When it comes time for budget cuts, you know where the cuts are made first. In my experience, of the company's professionals, the testers probably had the lowest standing in the company.

What is ironic here it that quality assurance testing is the only guarantee that a 'product' released works as it should. What is a common complaint you hear nearly every day? That the product you bought doesn't work as it should? That the software in the product doesn't work right? Or simply, that the product is awkward to use? Well, that's because no one wants to spend money or time on quality assurance (i.e. testing). A case in point is Micr*s*ft. The 'users' have become the 'testers'.

Cheers