[bug #54307] Wrong margins especially when printing one page

[Yavor Doganov]

URL:
<http://savannah.gnu.org/bugs/?54307>

Summary: Wrong margins especially when printing one page
Project: GNUstep
Submitted by: yavor
Submitted on: Fri 13 Jul 2018 06:45:26 PM EEST
Category: Gui/AppKit
Severity: 3 - Normal
Item Group: Bug
Status: None
Privacy: Public
Assigned to: None
Open/Closed: Open
Discussion Lock: Any

_______________________________________________________

Details:

Reported initially by Patrick Cardona in these threads:
http://lists.gnu.org/archive/html/discuss-gnustep/2018-07/msg00003.html
http://lists.gnu.org/archive/html/discuss-gnustep/2018-07/msg00005.html

There are several unrelated issues:

1. Some PPD files contain (valid) lines like the following:

*PrintQuality Draft/Draft:

This results in exception when parsing (example file at
http://lists.gnu.org/archive/html/discuss-gnustep/2018-07/msg00017.html).

2. In NSPageLayout -tableView:objectValueForTableColumn:row: the margins for
the standard papers are not correctly computed. imageRect already has the
printable margins but that is not taken into account so an expression like
paperSize.height - imageRect.size.height makes the topMargin twice larger than
it should be. Likewise for rightMargin. Also, the result is actually in pts
which gives funny figures; it should be converted to cm/in.

3. In NSPrintInfo -initWithDictionary: the margins should be set according to
ImageableArea as the FIXME comment says.

4. When a page with short text (few lines) has to be printed, the text does
not appear at the top of the page, as is expected, but is shifted way down
below. That's the most annoying item. There are some changes by Greg in the
"printing_fixes" branch which I think are related. I haven't tested them but
IMHO his approach is wrong and breaks printing of multiple pages (which
appears to be working correctly right now).

The attached patch fixes all 4 issues for me although I admit I'm not sure
whether it is correct. The whole repagination logic seems way too
awkward/convoluted to me and breaks reasonable WYSIWYG expectations.



_______________________________________________________

File Attachments:


-------------------------------------------------------
Date: Fri 13 Jul 2018 06:45:26 PM EEST Name:
0001-Miscellaneous-printing-fixes.patch Size: 4KiB By: yavor

<http://savannah.gnu.org/bugs/download.php?file_id=44553>

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Fred Kiefer]

Update of bug #54307 (project gnustep):

Status: None => Confirmed
Assigned to: None => FredKiefer

_______________________________________________________

Follow-up Comment #1:

Hi Yavor,

thank you for this bug report and the patch. I already had some similar
changes to NSPrinter implemented after the discussion on the mailing list.
That should be similar to your code but tries to unify different code paths.

As for NSPageLayout clearly the conversion was missing. I didn't like the way
you set the margins on both sides to the same values. So I just added the
corrected computation.

I also tried to add the same logic to the NSPrintInfo code.

Now with the NSPrintOperations change I understand which case you try to fix.
But the patch would break scaling and a lot of other stuff. And it does not
address the real issue here, just tries to work around in one special case. I
will try to experiment with the printing code a bit and see whether I can come
up woth something better.


_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Yavor Doganov]

Follow-up Comment #2, bug #54307 (project gnustep):

I thought the margins are always symmetrical, but anyway, your approach is
clearly better.

You are right that the NSPrintOperation change is just a dirty workaround and
apparently I haven't considered all subsequences. I hope some solution can be
found. Thanks for looking into this.

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Fred Kiefer]

Follow-up Comment #3, bug #54307 (project gnustep):

I tried very hard to understand a bit more of this issue. In the end I found
line 4503 in NSView.m:

if ((location.x != 0 || location.y != 0) && NSIsEmptyRect(aRect) == YES)

This only applies the following translation when the printed rectangle is
empty. If I remove the last part of the test, a single page print gets
positioned correctly. The question now is, why was this restriction here in
the first place and will it have any negative results on other printing
scenarios?

It really would help here, if you could do some experiments on your side as
well. First to confirm whether this resolves the specific issue and next on
negative side effects.

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Yavor Doganov]

Follow-up Comment #4, bug #54307 (project gnustep):

The last part of the test was removed by Eric Wasylishen on 23 December 2011
(commit d374f0a) precisely because of this problem. However, he reverted it
one hour later (commit 2f0f50c) with a comment that it broke Graphos
printing.

I have a CUPS server with two different printers configured. I can confirm
that with this change printing a single page works correctly for me with Ink,
TextEdit and Graphos. I don't know what Graphos breakage he had in mind back
then, perhaps he will remember? Or we can ask Riccardo to test as he is
undoubtedly more familiar with Graphos.

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Fred Kiefer]

Follow-up Comment #5, bug #54307 (project gnustep):

The check was originally introduced by Gregory Casamento on 13 Mar 2009. But
I don't expect that either of them (Eric and Greg) are following our
discussion here. Nor would they probably remember what they where doing back
then. We may have to figure this out ourselves :-(

If your try this change, please also do some multi page tests. I am almost
sure it will improve the single page case but we also need to inspect other
usages.

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Yavor Doganov]

Follow-up Comment #6, bug #54307 (project gnustep):

I also tested printing multiple pages, there is no change in the behavior with
Ink and TextEdit. I don't know how to do that with Graphos, though...

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

[Fred Kiefer]

Update of bug #54307 (project gnustep):

Status: Confirmed => Ready For Test
Open/Closed: Open => In Test

_______________________________________________________

Follow-up Comment #7:

I just committed the change to remove this is empty check before the
translation. This should correct the behaviour for half filled pages. There
surely are a lot more printing issues in GNUstep but the ones referenced in
this report should be fixed by this.

_______________________________________________________

Reply to this item at:

<http://savannah.gnu.org/bugs/?54307>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/