[texworks] external read-access causes TeXWorks to revert open file to old version

[Stefan Müller]

I'm absolutely not sure about this, as I've no experience with Qt, but this 
indeed sounds as if it could cause the trouble. Maybe at some place in TW this 
Qt  file caching mechanism isn't working as intended?

Stefan

Am Mittwoch 27 Oktober 2010, 07:18:28 schrieb Paul A Norman:
> Don't know what is being used  (where the trouble area might be), but
> looks like there are flushing procedures available in Qt files
> handling - don't know if this impinges on what has been happening or
> not ...
> 
>  From ... http://doc.trolltech.com/4.7/qfile.html#flush
> 
>   "  bool QFile::flush ()
>      Flushes any buffered data to the file. Returns true if successful;
>      otherwise returns false.  "
> 
> Hope this helps someone,
> 
> Paul
> 
> On 27 October 2010 12:15, Paul A Norman <paul.a.norman at gmail.com> wrote:
> > At least with QFileInfo caching is involved Qt 4.3 and up is this
> > affecting it?...
> > 
> > "To speed up performance, QFileInfo caches information about the file.
> > Because files can be changed by other users or programs, or even by
> > other parts of the same program, there is a function that refreshes
> > the file information: refresh(). If you want to switch off a
> > QFileInfo's caching and force it to access the file system every time
> > you request information from it call setCaching(false)."
> > 
> > http://doc.trolltech.com/4.3/qfileinfo.html
> > 
> > On 27 October 2010 07:55, Stefan Müller <warrence.stm at gmx.de> wrote:
> >> Hi again,
> >> 
> >> today I had a lot of commits in my document. I used Scite the whole time
> >> instead of TW without the described problem. I.e. no "positive run" with
> >> Scite. So it really may be a TW issue. It could have still been a
> >> coincidence, though...
> >> 
> >> Best wishes and thanks for looking at this issue,
> >> Stefan
> >> 
> >> On 26.10.2010 11:31, Paul A Norman wrote:
> >>> Thanks that clears a lot up Stefan.
> >>> 
> >>> Unless the same thing happens using another text editor - it is
> >>> outside of my experience area sorry.
> >>> 
> >>> Paul
> >>> 
> >>> On 26 October 2010 21:16, Stefan Müller<warrence.stm at gmx.de>  wrote:
> >>>> Hi Paul,
> >>>> 
> >>>> On 26.10.2010 02:12, Paul A Norman wrote:
> >>>>> Stefan,
> >>>>> 
> >>>>> Just to clarify from your earlier postings, you said that you have
> >>>>> two boxes, an Xp and a Linux box?
> >>>>> 
> >>>>> 1. Was this happening under both Os-es please? I'm still thinking of
> >>>>> caching issues here.
> >>>> 
> >>>> It happens under both Os-es. I'm using v0.3 r606 on Windows and v0.3
> >>>> r670 (with those extra debug lines from on of the previous mails) on
> >>>> Linux.
> >>>> 
> >>>>> 2. And have you tried exactly the same procedures of what you call a
> >>>>> "Positive run", not using TeXworks but instead a very simple text
> >>>>> editor (preferably a GUI one).
> >>>> 
> >>>> No, I haven't tried that. Mostly because I don't know any other nice
> >>>> TeX editor for Linux. I could try to reproduce a "positive run" with
> >>>> Scite. I'll
> >>>> report back soon about this.
> >>>> 
> >>>>> 3. Otherwise ... apparently there are two versions of TortoiseHG and
> >>>>> one of them is a Qt version -can you determine which one you are on?
> >>>>> ...
> >>>> 
> >>>> In the TortoiseHG info is something written about "Python, PyGTK,
> >>>> GTK", so I
> >>>> suppose I'm not using the Qt version. I didn't know there was one.
> >>>> 
> >>>> Stefan
> >>>> 
> >>>>> Thoughts ...
> >>>>> 
> >>>>> A. (i)(Starting to wonder if it is an intermittent Qt file handling
> >>>>> issue - does Qt use any memory caching in opening and accessing and
> >>>>> prior to commit to disk? Or build a virtual drive in Ram  for
> >>>>> temporary staging (same thing practically I suppose)?
> >>>>>     (ii) If so, is there an explicit flush/clear that needs to be
> >>>>> called?
> >>>>>     (iii) Is TortoiseHG Qt version accessing this same memory file
> >>>>> caching area (wouldn't
> >>>>>          that trigger some sort of a violation error message though
> >>>>> or does Qt leave a door
> >>>>>          open  for that between Qt applications?)?
> >>>>> 
> >>>>> Paul
> >>>>> 
> >>>>> On 26 October 2010 09:36, Stefan Müller<warrence.stm at gmx.de>    wrote:
> >>>>>> Another thing I just recognized: In a "positive run" (i.e. where the
> >>>>>> problem
> >>>>>> occurs) TW prints (beside others) the following debug messages:
> >>>>>> 
> >>>>>> Loading: "/dir/to/my/file.tex"
> >>>>>> Loading: "/dir/to/my"
> >>>>>> ...
> >>>>>> Loading: "/dir/to/my"
> >>>>>> Changed: "/dir/to/my/file.tex"
> >>>>>> Loading: "/dir/to/my/file.tex"
> >>>>>> 
> >>>>>> Is it correct, that sometimes the directory and sometimes the file
> >>>>>> is loaded?
> >>>>>> 
> >>>>>> Stefan
> >>>>>> 
> >>>>>> On 25.10.2010 19:19, Stefan Müller wrote:
> >>>>>>> Hi,
> >>>>>>> 
> >>>>>>> it has been a while since the last reply, but I was only now able
> >>>>>>> to test again on my Linux machine with those debug messages you
> >>>>>>> gave me (I
> >>>>>>> didn't compile it for Windows).
> >>>>>>> 
> >>>>>>> So, still strange things that happen. (Sometimes) When I change
> >>>>>>> something in the open tex file in TW, save and commit this change
> >>>>>>> with TortoiseHG (my Mercurial client), the debug messages say that
> >>>>>>> the open tex file has changed and that the same file is loaded
> >>>>>>> (again). Then TW displays an older version of the tex file. To get
> >>>>>>> my current file back I
> >>>>>>> have to reopen TW.
> >>>>>>> 
> >>>>>>> So there are basically two possiblities I can think of.
> >>>>>>> (1) TW has a problem with TortoiseHG accessing the tex file and
> >>>>>>> "restores" an old version of the file. What goes wrong? Where does
> >>>>>>> the old file come from?
> >>>>>>> (2) TortoiseHG messes with the tex file causing TW to reload it.
> >>>>>>> Why should TortoiseHG write on the file? And if TortoiseHG
> >>>>>>> temporarily writes to the tex file, why doesn't TW reloads again
> >>>>>>> when TortoiseHG restores the current (committed) version?
> >>>>>>> 
> >>>>>>> Hopefully someone has an idea what's causing the problem or how it
> >>>>>>> could
> >>>>>>> be fixed, especially if it's (1). Thanks in advance.
> >>>>>>> 
> >>>>>>> Best regards,
> >>>>>>> Stefan
> >>>>>>> 
> >>>>>>> On 22.09.2010 18:07, Stefan Löffler wrote:
> >>>>>>>> Am 2010-09-22 17:49, schrieb Stefan Müller:
> >>>>>>>>>> Anyway, if you build Tw yourself, you could add the following to
> >>>>>>>>>> lines
> >>>>>>>>>> (linenumber where to add in brackets) to get some debugging
> >>>>>>>>>> information:
> >>>>>>>>>> [959] qDebug()<<    "Loading:"<<    fileName;
> >>>>>>>>>> [1094] qDebug()<<    "Changed:"<<    curFile;
> >>>>>>>>>> This should print the filenames Tw loads, and whenever it
> >>>>>>>>>> detects a change to the filename on the harddisk.
> >>>>>>>>> 
> >>>>>>>>> Maybe that's not clever to ask, but in which file should I put
> >>>>>>>>> those lines? I
> >>>>>>>>> looked into TWApp.cpp, but it looks wrong there...
> >>>>>>>> 
> >>>>>>>> Oops, sorry. In src/TeXDocument.cpp.
> >>>>>>>> 
> >>>>>>>> Stefan