cupsConnectDest server-error-internal-error

[RolandHughes]

Ubuntu 20.04 all updates applied.
$ apt policy cups
cups:
Installed: 2.3.1-9ubuntu1.9
Candidate: 2.3.1-9ubuntu1.9
Version table:
*** 2.3.1-9ubuntu1.9 500
500 http://us.archive.ubuntu.com/ubuntu focal-updates/main amd64 Packages
500 http://security.ubuntu.com/ubuntu focal-security/main amd64 Packages
100 /var/lib/dpkg/status
2.3.1-9ubuntu1 500
500 http://us.archive.ubuntu.com/ubuntu focal/main amd64 Packages

Been slamming head against wall for over a week now. Maybe this has been fixed in later Cups versions, but if not, I want to send up a flare.
In the classic little example program

cupsEnumDests(CUPS_DEST_FLAGS_NONE, 1000, NULL, 0, 0, print_dest, NULL);

and print_dest can happily call this

		char resource[40000];
		const char *name = nullptr;
		char value[4096];
		http_t *http = cupsConnectDest(dest, CUPS_DEST_FLAGS_DEVICE, 30000, NULL, resource, sizeof(resource), NULL, NULL);

all is right with the world.

In a real program, when you obtain your dest pointer via

m_cupsDest = cupsGetNamedDest( CUPS_HTTP_DEFAULT, cupsGetDefault(), NULL);

if you try to use CUPS_DEST_FLAGS_DEVICE instead of CUPS_DEST_FLAGS_NONE you get an internal server error.

http_t *http = cupsConnectDest( m_cupsDest, CUPS_DEST_FLAGS_NONE, 30000, NULL, m_resource, RESOURCE_SIZE, NULL, NULL );

Why?

It shouldn't matter how one got the dest pointer should it? The structure is still the structure. It should still have all the same data.

Again, if this has been fixed, sorry for reporting but none of my searches could find any reports of this.

[michaelrsweet]

Closing this since the question references CUPS 2.x and was moved there.

[RolandHughes]

Actually, I'm trying to upgrade a C++ library to the 3.x API. The documentation is dramatically unclear as to what is and is not included with the 3.x API. The cups.h has IPP stuff in it. I've been in IT over 40 years now working on everything from mainframe to Midrange to medical devices and encrypted communications.

This is the doc I found.

It isn't that much different than the doc I found this morning after playing an awful lot of link tag.

Where is the 2.x place you moved this to? This stuff is so splayed out on the Internet finding the correct place to post or obtain documentation is nearly impossible.

cups-exp.zip

If you decompress that keeping the directory tree structure on a Ubuntu 20.04 (or any YABU of that base) having cmake and ninja along with C++ installed, you should be able to build the cups experiment example. It will create a build directory and an install directory, then install. It doesn't really "use" anything from LS-CS, but if the cmake pukes on that you can either nuke it from the cmake files for that project or install the current .deb. Sorry, that is compiled with full debug because I've been trying to track this down for a week.

in main.cpp in the print_dest() function you will see

http_t *http = cupsConnectDest(dest, CUPS_DEST_FLAGS_DEVICE, 30000, NULL, resource, sizeof(resource), NULL, NULL);

It works for every printer on my network as they all are IPP.

call that exact same function with a dest pointer generated by cupsGetNamedDest() which according to this documentation is valid in 3.x and it will fail despite having previously worked for the exact same printer.

I'm flagging this because something is obviously different with the content of the dest structure returned by the two calls.

Since you are the maintainer, please allow me to ask.

In the New World Order of Cups 3.x is all "printing" now supposed to be an application generated file that is queued to a printer?

I'm asking because Ls-Cs is a fork and gut of CopperSpice which is a fork and gut of Qt 4.8.x. There is a huge chunk of code, really many layers of classes, that is generating to a "pageinated device" printer. Basically relying on Cups or the physical printer's ability to consume and mostly bypassing concept of a print job. Under OS/2 and GUI-DOS (windows) we had to do this, especially if we needed our DOT-MATRIX to BOLD because we had to back up the print head and print again.

If "you generate the file and submit it as a job" is the new world order, please tell me because I have a lot of code to delete as part of this upgrade.

If the new world order of 3.x is currently supporting and will continue to support the philosophy of generating a page at a time for the "printer" requiring either Cups or the ipp printer to spool it until complete then please let me know that as well. Right now it isn't closing the job properly for the IPP world so the printer never gets the data.

I just want to do this right the first time instead of hacking a bandage on it which is far too common in the PC world.

Right now, the instant any Linux distro upgrades to 3.x only Cups all Qt based applications will cease to be able to print. That library uses the long obsolete PPD. I've got all of that purged. I got knee-capped by this dest pointer is not a dest pointer bug but now I must ask if there is more to nuke?

More to nuke would be a good thing. It would be a wonderful thing if the print dialog could have a "Content Type" option box providing

• generated PDF
• Raw binary
• Text

Then all supported operating systems would just queue the file. My text engine could even support page ranges counting lines and form feeds.

Thank you for your time.
Please answer.

[michaelrsweet]

@RolandHughes CUPS 2.x is the original CUPS repository, which is linked to from the OpenPrinting CUPS page.

CUPS 3.x is essentially a complete re-write for modern printers and printer applications. The CUPS library v3 is the basis of CUPS 3.x, with other projects providing a local spooler for general printing support and a sharing spooler for sharing printers, doing accounting, etc.

WRT your question:

In the New World Order of Cups 3.x is all "printing" now supposed to be an application generated file that is queued to a printer?

NO

The only time an application would connect to a printer would be to get status and capabilities (ready media, etc.) specific to the printer. However, since the capabilities of the print spooler are really what you want for a regular application, stick with CUPS_DEST_FLAGS_NONE when connecting and use the corresponding cupsCreateDestJob and cupsSendDestDocument functions to submit a job in a "standard" format like PDF.

[RolandHughes]

@michaelrsweet
Thank you for your reply.
When I look here. cupsSendDestDocument is not in that header file.

I think you mean for me to follow the Submitting A Print Job portion of the documentation. That is basically what I was asking.

We are talking past each other, mostly because I started out in computer operations on DEC/VAX equipment so all things in my brain are PRINT/QUEUE=queue_name/FORM=form_name file.name.

But yes, the printing can get rid of this nasty sending-page-at-a-time-while-connected pageination stuff. Just one engine to gen temp text file and one engine to generate temp pdf file. The temp is then chunked via

http_status_t
cupsStartDestDocument(http_t *http, cups_dest_t *dest,
                      cups_dinfo_t *info, int job_id,
                      const char *docname,
                      const char *format,
                      size_t num_options,
                      cups_option_t *options,
                      bool last_document);

http_status_t
cupsWriteRequestData(http_t *http, const char *buffer,
                     size_t length);

ipp_status_t
cupsFinishDestDocument(http_t *http, cups_dest_t *dest,
                       cups_dinfo_t *info);

only after creating a job

int job_id = 0;
size_ num_options = 0;
cups_option_t *options = NULL;

num_options = cupsAddIntegerOption(CUPS_COPIES, 42,
                                   num_options, &options);
num_options = cupsAddOption(CUPS_MEDIA, CUPS_MEDIA_LETTER,
                            num_options, &options);
num_options = cupsAddOption(CUPS_SIDES,
                            CUPS_SIDES_TWO_SIDED_PORTRAIT,
                            num_options, &options);

if (cupsCreateDestJob(CUPS_HTTP_DEFAULT, dest, info,
                      &job_id, "My Document", num_options,
                      options) == IPP_STATUS_OK)
  printf("Created job: %d\n", job_id);
else
  printf("Unable to create job: %s\n",
         cupsLastErrorString());

Thank you very much for your assistance.