Harlequin Core 14.0.r1 Release Notes
Introduction
Harlequin Core v14.0r1 is the first general maintenance release since v14.0r0. This release builds on v14.0r0 in that it addresses issues found by OEMs since release and includes performance improvements for Label and Packaging jobs.
Where we responded to customer issues the HQNSUP reference number is included in the "Harlequin Core v14.0r1 change details" table (below).
Jira Story | Support call(s) | Summary |
---|---|---|
HQN-386145 | HQNSUP-126908 | RIP can slow down as jobs progress Traversing the Pages tree of some Test jobs can slow on Scalable RIP. It's also slow in HHR if a PageRange is used because some jobs don't have a balanced tree as recommended by the PDF spec |
HQN-386140 | HQNSUP-127068 | HHR Crash with ACCESS VIOLATION when rendering multiple files Establish correct size for internal image buffers, fixing the crash |
HQN-386125 | n/a | Optimise createDLStores a Performance Core rip enhancement |
HQN-386121 | HQNSUP-127027 | Separation has invalid artefact in HHR14 A vector object was too large for the Rip size limits, fixed |
HQN-386095 | HQNSUP-127067 | Hamillroad screens not working with v14 HMR rip Registration of screening modules no longer checked for a previous definition, bugfix |
HQN-386094 | n/a | Add verify_system_requirements information to v14 Doxygen Discover the level of SIMD, SSE2, AVX2 supported by the CPU on which the rip is running is now documented in Doxygen. The means an update to the online api doc https://api.globalgraphics.com/harlequin/ |
HQN-386085 | HQNSUP-127033 | Crash in All separation with NextDevice transform This bug is a case where the rip shouldn't treat a color as a NextDeviceSpot when the rendering intent is None. This arose in a customer job CMYK + 3 spots + White |
HQN-386082 | HQNSUP-127038 | Job rips significantly faster on 13.0r1d than 13.2r4 or 14.0r0 Scan time had increased between 13.0r1d and 13.2r4 Optimise Usage of CMYKPassThrough With HVD |
HQN-386074 | HQNSUP-127016 | Using IndependentSpots gives unexpected results on Customer job |
HQN-385729 | n/a | Feature: HVD: Detect not touched hollow rectangles in strokes to improve HVD performance Development investigated the possibility of discovering a large untouched "hole" "in the middle of" a stroke. This is beneficial for some L&P jobs. We have examples of such jobs from several OEMs |
HQN-385727 | n/a | HVD Reuse: No first page reuse unless ignoring GTS/GGSL hints |
HQN-385966 | n/a | Ctrl-C stops working when -z is used to repeat the job The effect of using Ctl-C depended upon timing. Sometimes it very quickly shuts down all Farm RIPs, and returns to the cmd prompt. On other occasions, it did not |
HQN-386013 | n/a | Type 3 fonts didn't initialise some width fields giving bad output |
HQN-386106 | n/a | Scalable RIP can hang or crash if interrupted using Ctrl-C Affected all previous versions |
HQN-386065 | HQNSUP-127027 | Separation has invalid artefact in HHR14 Calculations for large smooth shfills may overflow and produce wrong output, bugfix of 64 bit rendering |
HQN-386031 | HQNSUP-126947 HQNSUP-126953 | TT Text Not Rendered 'Correctly' / Corrupted Text in Output Reduce cases of fix15 numeric overflow in TrueType code |
HQN-386014 | n/a | Non-zero overprints shouldn't be allowed in RGB transparency groups During internal testing we saw artifacts that resulted from CMYK overprints in an RGB transparency group |
HQN-385685 | SLD-661 | HVD: Allow the rip to accept the repetition period in the new OptimizedPDFRepetitionPeriod pdfparam |
Changes to supported platforms
Windows 11 Pro editions, version 22H2 will reach end of service on October 8, 2024
Add under Alerts and guidance
To enable modular screening in v14, the page setup now needs to set the Halftone (page device) parameter to /Modular
, whereas in v13, the required setting was true
. This distinction enables the RIP to avoid allocating unnecessary large band buffers, when a screening module is installed but not used.
Known issues
HQN-386054 - HVD + Tiling tmp files can exceed Ubuntu's open files limit, causing the RIP to error.
Occasionally element rasters that are being purged to disk can reach a cache limit on Ubuntu and cause the rip to error. At this time the Ubuntu terminal shows that it has a default limit of 1000 open files in the SW folder. Windows running the same job with 1644 tmp files causes no error. RedHat allows 1,000,000 and MacOS 10,240. Workaround for a single rip is to edit main.c as follows
https://api.globalgraphics.com/harlequin/main_8c.html
#include <sys/time.h>
#include <sys/resource.h>
int main(int argc, char *argv[])
{
int result = EXIT_SUCCESS ;
int i;
HqAssertHandlers_t assert_handlers = { HqCustomAssertMon, HqCustomTraceMon } ;
struct rlimit rl ;
rl.rlim_cur = 8192 ;
rl.rlim_max = 8192 ;
setrlimit(RLIMIT_NOFILE, &rl) ;
Documentation changes
Online documentation that were created or updated for Harlequin Core 14.0r1
Document | Status |
---|---|
updated | |
updated |