Hi, With some of the fonts (mainly Univers, CGTimes) some of the characters seems to be overlapping, only on Unix. whereas on windows everything seems to be fine. I could see following differences in the output. The Length in the FontDescriptor is less in case of Unix and the stream has some missing info at the end. ------------------------------------- Windows: /Length1 14036/Length 11284>>stream ..... Kf`4%1k6MJ5b1p*:]OL,=4MFq<Lm8mX]Euc=OkAo<Rk;Rc!OK,?e+OO<ZPUL/RDI@@+IbU<`NX1 9kRL4B@^p5<h3r*X]9lWFbfa,f"cb=?igBhJ$<ch!"]6dE*.gE=,#!SJ9W+?VuQg';#gT%"ZTfj `AE;lFTm:SL[YDthuj5Y"11,eY?b*l(k@,~> ------------------------------------- Unix: /Length1 14036/Length 11274>>stream ...... Kf`4%1k6MJ5b1p*:]OL,=4MFq<Lm8mX]Euc=OkAo<Rk;Rc!OK,?e+OO<ZPUL/RDI@@+IbU<`NX1 9kRL4B@^p5<h3r*X]9lZ(W)_ZJ0VtV75Fp'<s=C^"aF;Sp]lB.H@Pu!eIV]acr5YV4B"d[#>eM) G^f]VH[n@gJ3--%b4X:PnV@G~> ------------------------------------- I tried on AIX system with the following command: /hp/os/bin/gs -sDEVICE=ps2write -I/hp/os/lib/FONTS/Unicode/Resource - I/hp/os/lib/FONTS/Unicode -I/hp/os/lib/FONTS/Soft_Horizons -I/hp/os/lib/PS -q - dBATCH -dNOPAUSE -dSAFER -dSHORTERRORS -dWRITESYSTEMDICT -dGHOSTSCRIPT - dDEVICEWIDTHPOINTS=612 -dDEVICEHEIGHTPOINTS=792 -sOutputFile=out.ps -f input.ps The same issue is seen with pdfwrite device also. I guess the required fonts are already provided as part of other related bugs. I will attach the input postscript file and the output PS/PDF file. Thanks Subbu
Created attachment 4496 [details] input.ps simple input PS file ----- %!PS 15 500 moveto /Univers-BoldItalic 20.0 selectfont (ijk) show showpage -----
Created attachment 4497 [details] output_unix.pdf output_unix.pdf where we can see 'j' and 'k' are getting overlapped.
Created attachment 4498 [details] output_unix.ps output_unix.ps, here we can see the 'j' and 'k' are getting overlapped.
Created attachment 4499 [details] output_win.ps output_win.ps, this is the file generated on Windows, here we don't see any overlapping characters. everything seems to be fine in this.
Hi, Any update on this case? Thanks Subbu
This one is awaiting attention. I have been moved to a higher priority project for the forseeable future. If this is a problem, please contact Artifex to discuss this further.
Hi, We got a confirmation from our Font vendor that it is not a problem with the font and the issue could be that, GhostScript is not handling the placement of characters with a negative left side bearing correctly. In the example: 'j' seems to shift right about the same distance as the negative left side bearing. This is a composition, or character positioning problem. Thanks Subbu
Hi, Could you please take this defect up? Thanks Subbu
Hi, Any update on this? This is becoming critical now. Thanks Subbu
Hi, We are waiting for this fix. Thanks Subbu
Hi, Could you please take up this defect. Thanks Subbu
Please attach the Univers-BoldItalic font you are using in input.ps.
Hi, Please find the font in our following ftp site ftp://shkal:Patrons9@hprc.external.hp.com/ ftp://shkal:Patrons9@15.192.32.69/ Thanks subbu
I can't reproduce this. I've tried on Solaris 10, Ubuntu Linux 7.10 (on an amd64 system), and Mac OS X 10.5.5 with Ghostscript 8.62, 8.63, and head (r9246). All produce PDF files that Acrobat 8.0 and 9.0 read correctly (both display the output_unix.pdf file with the incorrect letter spacing). Either it's an AIX only issue or there is another copy of the Univers-BoldItalic font that is being used instead of the one referred to in comment #13. Also, are you sure you are running the same version of Ghostscript on Windows and AIX? The output_unix.ps file you attached is a PostScript Language Level 3 file, it starts with: %!PS-Adobe-3.0 whereas the output_win.ps file is a PostScript 2 file, it starts with: %!PS-Adobe-2.0 Finally, can you provide ssh access to an AIX system for testing?
Hi, Looks like this is AIX only issue. I don't think I can provide SSH access. I had given the difference I observed in the output file. THe difference you observed about %!PS-Adobe-2.0 is not a real problem, in case of unix we had an extra prologue but after that you had the same version below. Thanks Subbu
Hi, Just wanted to check if you don't have any AIX sytem to test this out? Thanks Subbu
We do not have access to an AIX system.
Would it be possible to check the difference in the output files and see what might be happening? Thanks Subbu
Hi Marcos, Since providing the access to our AIX system would be difficult, we are thinking of having "Virtual Room Desktop Share" setup and giving control to you for analysis. Please let me know if this okay for you and your convenient time for the same. Also one thing I observed is that, On AIX, with ps2write device there is no overlap in 8.54 version of GS, but with 8.63 it does. But with pdfwrite in both 8.54 and 8.63 I could see the overlap. And this issue is not there in solaris, linux, windows, everything is find here. Thanks Subbu
Is Virtual Room Desktop Share similar to using VNC? If so that would be acceptable; how about Thursday, 12/04/08 at 11:00 AM PST? If that time doesn't work I'm available anytime on Friday.
Hi Marcos, Yes, "Virtual Room Desktop Share" is similar to using VNC. Sicne 11:00 AM PST would be around 12:30 AM (midnight for me), it would be difficult for me and on Friday I will be on leave. So can we have it on Monday? Also can we have it during your late evening around 7:00PM (8.30 AM my time) or your early morning. Please let me know, I will send out the meeting request with the details. Thanks for your help. Subbu
Monday morning your time (i.e. Sunday night my time) would be fine. Can we do it at 10:00pm my time? What software will be I using on my end? Is there a way of testing it before Monday?
Hi Marcos, Sunday night 10PM your time, looks good to me. I will be sending out the URL for the virtual room, no extra softwares are required. Using this we can chat also. You can test your setup using this link https://www.rooms.hp.com/testsetup If you are using this for the first time it might install some of the required things. I will send out the meeting request with the dial-in numbers and the virtual room URL. Thanks Subbu
Created attachment 4647 [details] patch Proposed patch for this bug.
Thanks a lot for fix. One clarification, do we still need this line: if_debug2('l', "[l] ushort = %6u f = %6u\n", (ushort) (sbw[wmode] * factor), (ushort) f); Thanks Subbu
Created attachment 4648 [details] patch Lines that start with if_debug are for debugging and not needed. You also don't need the lines between the #if 0 and the #endif. I've attached a new, minimal patch.
Fixed by r9277.