(old)Puppy Linux Discussion Forum

Read here.

This problem also affects gnumeric. I tried reverting to the earlier libgnomeprint, but was unsuccessful.

Personally, I would rate this bug as a show-stopper.

This appears to be a CUPS 1.3.10 issue. I installed CUPS 1.3.10 into Puppy 4.1.2 and got the same printing problem (no printers detected) with Geany, Gnumeric, etc. I then recompiled Leafpad against the new CUPS and got the same problem.

I don't think that this is being caused by GTK.

I can't recall if CUPS 1.3.9 was OK so I will have to get a copy and try it next.

CUPS 1.3.9 is no better.

I don't understand why EPDF Viewer is able to see the CUPS printers when the other applications cannot. Was it compiled differently?

I had a look at NOP 3.14, which has CUPS 1.3.9. Unfortunately, all of the applications like Gnumeric have been removed, so it provides no clues.

rcrsn51 wrote:CUPS 1.3.9 is no better.

I don't understand why EPDF Viewer is able to see the CUPS printers when the other applications cannot. Was it compiled differently?

I had a look at NOP 3.14, which has CUPS 1.3.9. Unfortunately, all of the applications like Gnumeric have been removed, so it provides no clues.

Ok, I think the problem has to do with Geany and the postscript driver. If you do a print preview in Geany your printer will show up when you choose Print, but not when you choose print outside of Print Preview.

Testing 4.2.1RC3

Still have the same printing problem with Geany, Leafpad and Gnumeric - printers are not detected. Printing does work through Print Preview which calls ePDFView. Other applications work OK, including things like GtkLP.

As a test, I went back to 4.2.0 and dropped in CUPS 1.3.10 (both Patriot's and mine). Printing works fine.

My feeling now is that this is not being caused by CUPS per se. It is due to the inability of these three programs to connect to the cupsd daemon and extract printer information from it.

Hmmm .....

With regards to cups printing, there's a major difference between Geany and epdfview. Epdfview interfaces directly to cups library for printing support while Geany relies on GTK+ to provide printing support. I'll try to see if fresh gtk+ binaries on newer cups have any impact.

Rgds

Patriot wrote:With regards to cups printing, there's a major difference between Geany and epdfview. Epdfview interfaces directly to cups library for printing support while Geany relies on GTK+ to provide printing support. I'll try to see if fresh gtk+ binaries on newer cups have any impact.

....and Gtk+ was changed from 4.2 to 4.2.1 to fix an xinput issue.

The other change I made that seems to make a difference is installing Liberation truetype fonts. They were too big for the standard release but I've included them for uploading to the repository. Could simply having the appropriate truetype fonts available make it possible for Geany to connect via Gtk+?

Hmmm .....

Ok, I've made fresh binaries of Gtk+ 2.12.12 (based on Pup412) ... Preliminary tests with geany, leafpad, abiword & gnumerics shows that cups (1.3.10) printer is listed and selectable for printing.

I will need to do more tests later to determine some other stuffs. It's been a long day ...

Rgds

Patriot wrote:Hmmm .....

Ok, I've made fresh binaries of Gtk+ 2.12.12 (based on Pup412) ... Preliminary tests with geany, leafpad, abiword & gnumerics shows that cups (1.3.10) printer is listed and selectable for printing.

I will need to do more tests later to determine some other stuffs. It's been a long day ...

The reason Gtk+ was patched was to allow SCIM to function. Pizzasgood did the patching, although there was a version by ttuuxxx as well. It's PG's version in 4.2.1 though.

It might be worth checking with PG to see what he needed to patch to get SCIM to function.

OK.

I went to Puppy 4.2.0 and got the folder /usr/lib/gtk-2.0/2.10.0/printbackends.

I then went to Puppy 4.2.1 RC3 and replaced its printbackends folder with the one from 4.2.0.

Printers are now detected in Geany, Gnumeric, etc. Other apps like Abiword still print OK.

How might this regression affect SCIM?

[Edit] As a test, I then went back to Puppy 4.1.2 and installed CUPS 3.1.10. In this scenario, Gnumeric cannot detect printers, supposedly because its GTK+ was compiled against the older CUPS 1.1.23.

I then dropped in the backends folder from 4.2.0 and printing was fixed!

I suspect that in 4.2.0, GTK+ was compiled against CUPS 1.4b2, which is close enough to 1.3.10 that the backend works in both versions.

Hmmm .....

WhoDo wrote:.....
It might be worth checking with PG to see what he needed to patch to get SCIM to function.

I remembered reading about enabling debug and adding Xinput. Problem is, I don't have any scim hardware. Maybe I'll make another set later with the extra options and see how it goes.

rcrsn51 wrote:.....
I went to Puppy 4.2.0 and got the folder /usr/lib/gtk-2.0/2.10.0/printbackends. .....

I second that finding. Was among the things I replaced/mixed-in. I have no idea if it will break scim.

Rgds

So is this just a matter of recompiling the patched GTK+ against the right version of CUPS?

rcrsn51 wrote:So is this just a matter of recompiling the patched GTK+ against the right version of CUPS?

I believe so, yes. I recalled some internal differences between cups 1.1.x and 1.3.x (and no, I didn't go through it all yet) and it must've escaped me with gtk+

That would mean some core stuffs that relies on cups libs interface may need to be checked (I'll have to check my samba again). Hopefully the ones using gtk+ printing interface won't have to be.

Rgds

Hmmm .....

Ok, I've tested another set of gtk+ 2.12.12 binaries with xinput enabled (indicated on PG's post ) against cups 1.3.10 ... Gtk+ apps printers are listed and selectable. Pressing the insert key in text input boxes does not crash Seamonkey. So, I guess that confirms the need for Gtk+ to be compiled against intended cups version.

Right now, I'll just stick to Gtk+ 2.12.12 and see how it goes ...

Rgds