| <<<Back 1 day (to 2022/04/14) | Fwd 1 day (to 2022/04/16) >>> | 20220415 |
artifexirc-bot | <Kotyara_Single_Cat> Hi! I found in **configure** file --without-tesseract flag — it's help me on AIX5.3. But on AIX7.1 with ghostscript-9.55.0 I have new error: | 07:06.14 |
| <Kotyara_Single_Cat> "./base/gxpcmap.c", line 1509.13: 1506-046 (S) Syntax error. | 07:06.15 |
| <Kotyara_Single_Cat> "./base/gxpcmap.c", line 1511.18: 1506-045 (S) Undeclared identifier padev. | 07:06.16 |
| <Kotyara_Single_Cat> "./base/gxpcmap.c", line 1595.9: 1506-046 (S) Syntax error. | 07:06.18 |
| <Kotyara_Single_Cat> gmake: *** [base/lib.mak:2653: obj/gxpcmap.o] Error 1 | 07:06.19 |
| <KenSharp> Well if it compiles on AIX 5.3 but not 7.1 then that shounds like a compiler difference to me. | 07:08.02 |
| <chrisl> There isn't a "padev" identifier used around those line numbers | 07:09.27 |
| <Kotyara_Single_Cat> I found this topic: https://community.ibm.com/community/user/power/communities/community-home/digestviewer/viewthread?GroupId=6211&MID=141097&CommunityKey=10c1d831-47ee-4d92-a138-b03f7896f7c9&ReturnUrl=%2Fcommunity%2Fuser%2Fcommunities%2Fcommunity-home%2Fdigestviewer%3FCommunityKey%3D10c1d831-47ee-4d92-a138-b03f7896f7c9 | 07:09.39 |
| <Kotyara_Single_Cat> And tried to use 9.53.3 ghostscript — error reproduce. | 07:09.39 |
| <Kotyara_Single_Cat> https://cdn.discordapp.com/attachments/773567375458828329/964422234963865620/unknown.png | 07:09.40 |
| <KenSharp> Thanks for checking Chris, I haven't had time yet | 07:09.52 |
| <KenSharp> @chrisl there is a padev in the 9.55.0 sources at that location | 07:17.27 |
| <KenSharp> Line 1509 is the declaration: | 07:17.58 |
| <KenSharp> gx_device_pattern_accum *padev = (gx_device_pattern_accum *) adev; | 07:17.58 |
| <KenSharp> SOrry 1511, not1509 | 07:18.07 |
| <chrisl> Oh, okay... | 07:18.10 |
| <chrisl> It looks pretty "declared" to me! | 07:18.26 |
| <KenSharp> It is, I'm just wondering if the .h file is explicitly included | 07:18.42 |
| <KenSharp> needs gxpcolor.h | 07:18.56 |
| <KenSharp> And yes, that's included in the head of the .c file | 07:19.13 |
| <chrisl> And any compiler ought to fail, if that were missing | 07:19.26 |
| <KenSharp> Oh.... | 07:19.54 |
| <KenSharp> Someone has added a C++ comment! | 07:20.03 |
| <KenSharp> That's almost certainly what it's complaining about | 07:20.13 |
| <chrisl> There are C++ comments dotted around | 07:20.24 |
| <chrisl> Including in pdfi 😦 | 07:20.29 |
| <KenSharp> @Kotyara_Single_Cat try changing : | 07:20.44 |
| <KenSharp> // free pattern cache data that never got added to the dictionary | 07:20.45 |
| <KenSharp> To : | 07:20.46 |
| <KenSharp> /* free pattern cache data that never got added to the dictionary */ | 07:20.47 |
| <KenSharp> Yeah I've been meaning to go back and remove those. I used them in a couple of places as reminders that the comment needed removed | 07:21.14 |
| <chrisl> lcms is also full of C++ comments | 07:22.13 |
| <KenSharp> Well if that is the problem (and that's line 1509 in my source here) then @Kotyara_Single_Cat is going to have to find another compiler to use or a way to make it accept C++ comments. | 07:22.58 |
| <chrisl> I wonder how the hell I find the xlc version.... | 07:24.02 |
| <KenSharp> You're logged into compile farm ? | 07:24.14 |
| <chrisl> Yeh | 07:24.20 |
| <KenSharp> -qversion | 07:24.37 |
| <KenSharp> according to IBM | 07:24.41 |
| <chrisl> IBM XL C/C++ for AIX, V12.1 (5765-J02, 5725-C72) | 07:25.29 |
| <chrisl> Version: 12.01.0000.0000 | 07:25.29 |
| <chrisl> Going to try gcc first, though | 07:25.45 |
| <KenSharp> Fair enough, but the fact that it says C/C++ suggests it can cope with C++ comments.... | 07:26.04 |
| <KenSharp> @chrisl FWIW line 1509 in the 9.55.0 sources is line 1548 in yesterdays' HEAD. I haven't picked up Robin and Michael's commits yet | 07:27.42 |
| <Kotyara_Single_Cat> I star compilation. | 07:27.46 |
| <Kotyara_Single_Cat> What surprises me the most is the fact that everything works well on 5.3 without any changes in code... | 07:28.52 |
| <KenSharp> Some kind of compiler difference would seem the most likely. You might like to compare the compiler versions | 07:29.15 |
| <Kotyara_Single_Cat> AIX5.3: | 07:33.35 |
| <Kotyara_Single_Cat> /usr/vacpp/bin/xlC -qversion | 07:33.36 |
| <Kotyara_Single_Cat> IBM XL C/C++ Enterprise Edition for AIX, V9.0 | 07:33.37 |
| <Kotyara_Single_Cat> Version: 09.00.0000.0007 | 07:33.38 |
| <Kotyara_Single_Cat> AIX7.1: | 07:33.39 |
| <Kotyara_Single_Cat> /usr/vacpp/bin/xlC -qversion | 07:33.41 |
| <Kotyara_Single_Cat> IBM XL C/C++ Enterprise Edition for AIX, V9.0 | 07:33.42 |
| <Kotyara_Single_Cat> Version: 09.00.0000.0019 | 07:33.43 |
| <Kotyara_Single_Cat> AIX5.3: | 07:33.46 |
| <Kotyara_Single_Cat> /usr/vacpp/bin/xlC -qversion | 07:33.47 |
| <Kotyara_Single_Cat> IBM XL C/C++ Enterprise Edition for AIX, V9.0 | 07:33.49 |
| <Kotyara_Single_Cat> Version: **09.00.0000.0007** | 07:33.50 |
| <Kotyara_Single_Cat> AIX7.1: | 07:33.51 |
| <Kotyara_Single_Cat> /usr/vacpp/bin/xlC -qversion | 07:33.52 |
| <Kotyara_Single_Cat> IBM XL C/C++ Enterprise Edition for AIX, V9.0 | 07:33.54 |
| <Kotyara_Single_Cat> Version: **09.00.0000.0019** | 07:33.55 |
| <KenSharp> Well it's a very small difference, but I'm not familiar enough with IBM compilers to know if it's significant | 07:34.10 |
| <chrisl> @Kotyara_Single_Cat With gcc 8.3.0, "./configure --without-tesseract && make -j4" just ran to completion. with a small tweak to the sources under "extract/". And I get a working binary from it | 07:36.04 |
| <chrisl> The tweak to extract/ is because the compiler is over fussy about a function it thinks doesn't have a proper "return" at its end | 07:37.22 |
| <chrisl> Trying xlc now.... | 07:37.53 |
| <Kotyara_Single_Cat> Didn't help | 07:37.59 |
| <KenSharp> Then I don't know what the compiler is complaining about. | 07:38.30 |
| <chrisl> FWIW, I'm accessing AIX 7.1.5 | 07:38.45 |
| <KenSharp> You oculd try removing the comment altogether | 07:38.46 |
| <chrisl> Going to get a coffee... back shortly | 07:40.14 |
| <Kotyara_Single_Cat> Me too. To be honest I don't know C++ and I apologise if my messages stupid. Maybe full log can help? | 07:45.39 |
| <Kotyara_Single_Cat> https://cdn.discordapp.com/attachments/773567375458828329/964431295608414208/logs.txt | 07:45.40 |
| <chrisl> Tesseract is the only C++ involved, which we're leaving out using the configure option | 07:46.47 |
| <chrisl> Hmm, it's not actually calling xlc, it's calling "cc".... | 07:47.46 |
| <Kotyara_Single_Cat> Hmmm, I'll try to find "cc" version | 07:48.36 |
| <chrisl> "cc" is probably a symlink, causing xlc to run in a specific configuration. | 07:49.11 |
| <chrisl> So, xlc *does* fail to build for me, but not in the way you are seeing | 07:49.37 |
| <chrisl> Oh, it looks like it's because memcmp is a macro <sigh> | 07:51.38 |
| <chrisl> @Kotyara_Single_Cat So... gcc 8.3.0 on AIX 7.1.5 works for me - after, as I said, a very small tweak to the extract/ code | 07:52.57 |
| <Kotyara_Single_Cat> You right. __AIX5.3:__ 9.0.0.7, __AIX7.1:__ 9.0.0.19 | 08:14.01 |
| <Kotyara_Single_Cat> Oh, I found. AIX5.3 used gcc 8.3.0 compiler, AIX7.1 build with xLC 9.0.0.19 | 08:17.47 |
| <Kotyara_Single_Cat> @chrisl @KenSharp Thank you very much for help! I think using gcc on AIX7.1 should fix the problem | 08:17.48 |
| <Kotyara_Single_Cat> Oh, I found. AIX5.3 used gcc 8.3.0 compiler, AIX7.1 build with xLC 9.0.0.19 | 08:18.03 |
| <Kotyara_Single_Cat> @chrisl @KenSharp Thank you very much for help! I think using gcc on AIX7.1 should fix the problem 🙂 | 08:18.03 |
| <KenSharp> Certainly using gcc is a better option from our point of view, since that is what we mainly support. Other C compilers shuold work of course, but there can be difficulties | 08:18.34 |
| <chrisl> I will make a note to sort out the the problems that I can see with xlc | 08:19.18 |
| <<<Back 1 day (to 2022/04/14) | Forward 1 day (to 2022/04/16)>>> | |