ghostscript: update to 10.01.0

[remkos]

Description

Update to upstream version 10.01.0.

From the developers: This release officially deprecates the old Postscript implementation of PDF. The option to use the old PDF implementation will be removed in the next full release (10.01.0).

This also removes an unused patch file.

In addition, a revision bump for xfig is included to make sure this new version of the library is linked.

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 13.4.1 22F770820d x86_64
Command Line Tools 14.3.1.0.1.1683849156

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL?
• checked your Portfile with port lint --nitpick?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?
• checked that the Portfile's most important variants haven't been broken?

[tomio-arisaka]

Ghostscript 10.01.1 is released.

From the developers: This release officially deprecates the old Postscript implementation of PDF. The option to use the old PDF implementation will be removed in the next full release (10.01.0).

On Ghostscript 10.01.1, -dNEWPDF=false is removed, but the PDF interpreter written in PostScript is still available. You can see it if you check the differences between Ghostscript 10.00.0 and 10.01.1.

Besides, I read some serious bug reports on the 10.01.0 release, not on 10.0.0.

What is the serious bug?
Is it a vulnerability shown on the next web page?
https://nvd.nist.gov/vuln/detail/CVE-2023-28879
If so, it is fixed on Ghostscript 10.01.1.
https://git.ghostscript.com/?p=ghostpdl.git;a=shortlog;h=refs/heads/gs10.01.1

[remkos]

I read of issues of rendering certain PS into PNG images, seemingly totally uncorrelated to the PDF interpreter.
Here is the reference: GenericMappingTools/gmt#7336

[tomio-arisaka]

I read of issues of rendering certain PS into PNG images, seemingly totally uncorrelated to the PDF interpreter. Here is the reference: GenericMappingTools/gmt#7336

Thanks for the info.

I have not ever used GMT.
But I think the issue occurs due to using pngalpha or x11alpha.
('alpha' means that the background is transparent)
For example, GMT uses the pngalpha device as follows:

$ git clone https://github.com/GenericMappingTools/gmt.git gmt-code
Cloning into 'gmt-code'...
remote: Enumerating objects: 279730, done.
remote: Counting objects: 100% (1256/1256), done.
remote: Compressing objects: 100% (594/594), done.
remote: Total 279730 (delta 765), reused 1082 (delta 654), pack-reused 278474
Receiving objects: 100% (279730/279730), 875.04 MiB | 10.22 MiB/s, done.
Resolving deltas: 100% (208976/208976), done.
$ 
$ find ./gmt-code -type f -exec grep 'pngalpha' {} +
./gmt-code/admin/gs_check.sh:  -sDEVICE=pngalpha  -g2550x3300 -r300 -sOutputFile='PS_orig_for_gs.png' 'PS_orig_for_gs_intermediate.pdf'
./gmt-code/src/psconvert.c:	char *device[N_GS_DEVICES] = {"", "pdfwrite", "svg", "jpeg", "png16m", "ppmraw", "tiff24nc", "bmp16m", "pngalpha",
./gmt-code/src/psconvert.c:		"", /* pngalpha */
$ 

So I made a patch to resolve the issue of Ghostscript-10.01.1.

--- A/base/gspaint.c	2023-03-27 15:49:35.000000000 +0900
+++ B/base/gspaint.c	2023-04-27 19:13:03.000000000 +0900
@@ -208,6 +208,7 @@
     gs_fixed_rect bbox;
     gs_int_rect ibox;
     uint width, raster, band_space;
+    uint dev_width, dev_height;
     uint height, height2;
     gs_log2_scale_point log2_scale;
     gs_memory_t *mem;
@@ -223,6 +224,7 @@
     if ((ibox.q.y <= ibox.p.y) || (ibox.q.x <= ibox.p.x))
         return 2;
     width = (ibox.q.x - ibox.p.x) << log2_scale.x;
+    dev_width = ibox.q.x << log2_scale.x;
     raster = bitmap_raster(width);
     band_space = raster << log2_scale.y;
     height2 = (ibox.q.y - ibox.p.y);
@@ -232,6 +234,7 @@
     if (height > height2)
         height = height2;
     height <<= log2_scale.y;
+    dev_height = ibox.q.y << log2_scale.y;
     mem = pgs->memory;
     mdev = gs_alloc_struct(mem, gx_device_memory, &st_device_memory,
                            "alpha_buffer_init");
@@ -244,13 +247,9 @@
     }
     gs_make_mem_abuf_device(mdev, mem, dev, &log2_scale,
                             alpha_bits, ibox.p.x << log2_scale.x, devn);
-    mdev->width = width;
-    mdev->height = height;
+    mdev->width = dev_width;
+    mdev->height = dev_height;
     mdev->bitmap_memory = mem;
-    /* Set the horrible hacky flag that tells people that the width/height here
-     * have been set for *our* convenience, rather than accurately depicting the
-     * size of the device for callers. */
-    mdev->non_strict_bounds = 1;
     if ((*dev_proc(mdev, open_device)) ((gx_device *) mdev) < 0) {
         /* No room for bits, punt. */
         gs_free_object(mem, mdev, "alpha_buffer_init");

This patch is based on the next commit.
https://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=c52b4a80cdfc6d9b7d04f52c4a872717aeeff73c

Are you able to check this patch?

[remkos]

@tomio-arisaka : Your patch indeed works to resolve what was seen in GenericMappingTools/gmt#7336 with ghostscript v10.01.0.

I've applied your patch to ghostscript v10.01.2 and that removed the artifacts seen.
Therefore, I've created a new PR #19510

Thanks if you can then approve it.

[remkos]

This PR is now superseded by PR #19510
@tomio-arisaka : Please review that one.