[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Texmacs-dev] Building TeXmacs for macos

From: Pierre-Henri Jondot
Subject: Re: [Texmacs-dev] Building TeXmacs for macos
Date: Sat, 14 Nov 2020 17:18:27 +0100

Although it is indeed not the same problem, I can confirm the bug you mention 
here. With texmacs 1.99.15 built with qt 5.15.1, texmacs segfaults opening any 
session but scheme, with the build I obtain on macos catalina. No such problem 
with qt 4 (and the build is easier with qt 4…). A python session doesn’t work, 
but at least it doesn’t crash...

So the problem probably is not related to freebsd but indeed with qt.


P.S. : could someone try to :
- make sure texmacs is configured to use multiple windows (Edit-Preferences 
last option)
- initiate a drawing
- while still in drawing mode, open another window…
Am I the only for whom this makes texmacs crash ? (I doubt it, but so far…)

> Le 14 nov. 2020 à 16:23, Kostas Oikonomou <k.oikonomou@att.net> a écrit :
> Yes, I understand his problem is at build time.  I just wanted to point out 
> that on another, perhaps somewhat similar,  OS there is no such problem.
> I also forgot to add that
> 1) The compiler used is clang 10 (standard on FreeBSD)
> 2) TeXmacs 1.99.{13,14} not only builds, but also works fine, with the 
> exception of plugins (pipes).
> 3) The situation with 1.99.15 is the same: TeXmacs builds, works, but not 
> with plugins.
>                             Kostas
> On 11/14/20 9:51 AM, Massimiliano Gubinelli wrote:
>> Dear Kostas,
>>  as far as I understand the problems of Pierre-Henri is at buildtime, not 
>> runtime. It seems to me they are different. Anyway it would be better to 
>> move the discussion to texmacs-dev.
>> Max
>>> On 14. Nov 2020, at 15:29, Kostas Oikonomou <k.oikonomou@att.net> wrote:
>>> Just to add to the Qt5 discussion (confusion?), as I wrote a few days 
>>> earlier, this is what happens with FreeBSD.
>>> There are no problems compiling with Qt5, but
>>>> Can someone say with which version of Qt5 is TeXmacs 1.99.14 (or 1.99.13, 
>>>> for that matter) supposed to work?
>>>> I have Qt 5.15 on FreeBSD, and both 1.99.13 and 1.99.14 segfault when I 
>>>> try to use *any* plugin (except Scheme).
>>>> GDB shows that the problem has to do with libQt5Core, more specifically 
>>>> with the use of Qt pipes. This problem did not occur with Qt 5.14.
>>>>                     Kostas
>>> and then
>>>> In any case, I downgraded Qt to 5.14, recompiled, and now plugins work in 
>>>> 1.99.14.
>>>> So something seems to have changed in Qt 5.15 with respect to pipes, at 
>>>> least for FreeBSD.
>>>>                 Kostas
>>> On 11/14/20 9:07 AM, Massimiliano Gubinelli wrote:
>>>> I can add that building with Qt5 works fine in my machine and as far as I 
>>>> know also for Joris and Denis (always on Mac). So there should be 
>>>> something different going on on your specific setup.
>>>>> On 14. Nov 2020, at 15:04, Massimiliano Gubinelli <m.gubinelli@gmail.com> 
>>>>> wrote:
>>>>> The -isysroot option should come from Qt, gathered by configure from the 
>>>>> qmake output. Probably Qt5 was build with 10.11 sdk. I'm not sure. 
>>>>> However independently of what is your running OS I was meaning that the 
>>>>> headers of Big Sur should have changed, because there is no such problems 
>>>>> with SDKs from previous systems (e.g. Catilina).
>>>>> To investigate the issue you can use a very simple Qt application and try 
>>>>> to build it with qmake to see which headers are used. As I said, the 
>>>>> error messages you get seems to come from changes in the headers which 
>>>>> cause symbol clashes in TeXmacs. We had this kind of issues before due to 
>>>>> the fact that various libraries might have incompatible #define
>>>>> Max
>>>>>> On 14. Nov 2020, at 14:45, Pierre-Henri Jondot <phjondot@gmail.com> 
>>>>>> wrote:
>>>>>> Again : I am still running catalina too… but it seems that the configure 
>>>>>> step when building with qt 5 had problems… configure chose to build with 
>>>>>> 10.11 sdk (with -isysroot directive) by default which fails. It is 
>>>>>> possible to build with qt 5, but for that to succeed, I had to change 
>>>>>> quite a few paths in makefile.
>>>>>> I haven’t explored Darcy Shen suggestion to use CMake instead yet, but 
>>>>>> it might be another way out.
>>>>>> Regards,
>>>>>> Pierre-Henri
>>>>>>> Le 14 nov. 2020 à 14:31, Massimiliano Gubinelli <m.gubinelli@gmail.com> 
>>>>>>> a écrit :
>>>>>>> I might add that on Catilina (MacOS 10.15) TeXMacs compile fine with Qt 
>>>>>>> 5 (e.g. 5.9 or 5.15). Maybe something is changed in Big Sur (MacOS 11) 
>>>>>>> at the level of the include files and this makes compilation fails. For 
>>>>>>> the moment I do not want to update my machine so I cannot indagate 
>>>>>>> further.
>>>>>>> Max
>>>>>>>> On 14. Nov 2020, at 13:28, Pierre-Henri Jondot <phjondot@gmail.com> 
>>>>>>>> wrote:
>>>>>>>> Hi,
>>>>>>>> We exchanged a few mails with Max, and the conclusion is that the 
>>>>>>>> problems I had were the result of trying to build TeXmacs with qt5, 
>>>>>>>> which is possible it seems, although not yet supported and a bit of a 
>>>>>>>> pain, as it seems to really confuse the configure step. I think the 
>>>>>>>> instructions on the website regarding compiling TeXmacs from the 
>>>>>>>> sources should be edited to advice compiling with Qt4 and not Qt5. (No 
>>>>>>>> version is suggested there, which led me to believe the current 
>>>>>>>> version of Qt was the expected one, or at least as good as the 
>>>>>>>> previous one…)
>>>>>>>> Regards,
>>>>>>>> Pierre-Henri
>>>>>>>>> Le 14 nov. 2020 à 09:03, Massimiliano Gubinelli 
>>>>>>>>> <m.gubinelli@gmail.com> a écrit :
>>>>>>>>> That is strange. Building on mac is usually ok for me, with homebrew 
>>>>>>>>> for example. Which Qt do you use? Can you post the compilation 
>>>>>>>>> invocation for mac_app.mm? (some few lines before the error message)
>>>>>>>>> Max

reply via email to

[Prev in Thread] Current Thread [Next in Thread]