I noticed a shQuote fix for Windows in the svn logs.
Just wanted to point out that this will favorably
affect texi2dvi on Windows which previously used
UNIX quoting and so generated an incorrect Windows
command. (Note that texi2dvi is used when creating
vignettes.)
Another problem is that the recommended tex
distribution for Windows, fptex, does not have texi2dvi
in the first place. The alternative, MiKTeX, is harder
to install than fptex but instructions are available
and it does have an equivalent to texi2dvi called texify.exe;
however, R still does not know about texify.exe
resulting in additional installation hassles,
viz. setting the texi2dvi option or setting up a texi2dvi.bat
file yourself that calls texify.exe or just forgetting
about texi2dvi and manually running the necessary tex commands
when creating vignettes on Windows.
There are also additional problems related to using the
perl package-building scripts in Windows. I can create the
zoo vignette using:
setwd("...whatever...")
Sweave("zoo.Rnw")
texi2dvi("zoo.tex")
provided I modify the source to texi2dvi to use the
appropriate quoting but even with that modification,
the perl package build scripts only build
the vignette as far as the zoo.tex file and I must
do the rest by hand so there is some problem with
the scripts or in how they interact with R.
I think the real ultimate solution is to replace the
perl scripts with R code to make them more maintainable.
This seems like the direction things are moving anyways
and I suspect that they were developed long before R
got to its current advanced state where R is just
as powerful as perl.
Gabor Grothendieck wrote:> I noticed a shQuote fix for Windows in the svn logs. > Just wanted to point out that this will favorably > affect texi2dvi on Windows which previously used > UNIX quoting and so generated an incorrect Windows > command. (Note that texi2dvi is used when creating > vignettes.) > > Another problem is that the recommended tex > distribution for Windows, fptex, does not have texi2dvi > in the first place. The alternative, MiKTeX, is harder > to installI heared some people think so, but apparently it is not a big problem. > than fptex but instructions are available> and it does have an equivalent to texi2dvi called texify.exe; > however, R still does not know about texify.exe > resulting in additional installation hassles, > viz. setting the texi2dvi option or setting up a texi2dvi.bat > file yourself that calls texify.exe or just forgetting > about texi2dvi and manually running the necessary tex commands > when creating vignettes on Windows.texi2dvi is also available in MikTeX, see package miktex-texinfo-bin. See also ?texi2dvi in R to see how you can set something different, e.g. "texify", which is explicitly given in the examples.> There are also additional problems related to using the > perl package-building scripts in Windows. I can create the > zoo vignette using: > > setwd("...whatever...") > Sweave("zoo.Rnw") > texi2dvi("zoo.tex") > > provided I modify the source to texi2dvi to use the > appropriate quoting but even with that modification, > the perl package build scripts only build > the vignette as far as the zoo.tex file and I must > do the rest by hand so there is some problem with > the scripts or in how they interact with R.Are you going to provide a patch? Uwe Ligges> I think the real ultimate solution is to replace the > perl scripts with R code to make them more maintainable. > This seems like the direction things are moving anyways > and I suspect that they were developed long before R > got to its current advanced state where R is just > as powerful as perl. > > ______________________________________________ > R-devel@stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel
From: Uwe Ligges <ligges@statistik.uni-dortmund.de>> Gabor Grothendieck wrote: > > > I noticed a shQuote fix for Windows in the svn logs. > > Just wanted to point out that this will favorably > > affect texi2dvi on Windows which previously used > > UNIX quoting and so generated an incorrect Windows > > command. (Note that texi2dvi is used when creating > > vignettes.) > > > > Another problem is that the recommended tex > > distribution for Windows, fptex, does not have texi2dvi > > in the first place. The alternative, MiKTeX, is harder > > to install > > I heared some people think so, but apparently it is not a big problem.I installed both so I know from experience that MikTex was harder to install. The problem is that its a lot of work to set up a package building environment on Windows. There are numerous systems to install, TeX, UNIX tools, perl, Microsoft Help plus path problems. MikTex is just one more obstacle. MikTex seems like its a slicker distribution package so its too bad its more work to set up. The main problems with fptex are not configuration but that (1) the package is so large that it takes a long time to download and (2) its missing crucial elements, viz. texi2dvi.> > > > than fptex but instructions are available > > and it does have an equivalent to texi2dvi called texify.exe; > > however, R still does not know about texify.exe > > resulting in additional installation hassles, > > viz. setting the texi2dvi option or setting up a texi2dvi.bat > > file yourself that calls texify.exe or just forgetting > > about texi2dvi and manually running the necessary tex commands > > when creating vignettes on Windows. > > texi2dvi is also available in MikTeX, see package miktex-texinfo-bin. > > See also ?texi2dvi in R to see how you can set something different, e.g. > "texify", which is explicitly given in the examples. > > > > > There are also additional problems related to using the > > perl package-building scripts in Windows. I can create the > > zoo vignette using: > > > > setwd("...whatever...") > > Sweave("zoo.Rnw") > > texi2dvi("zoo.tex") > > > > provided I modify the source to texi2dvi to use the > > appropriate quoting but even with that modification, > > the perl package build scripts only build > > the vignette as far as the zoo.tex file and I must > > do the rest by hand so there is some problem with > > the scripts or in how they interact with R. > > Are you going to provide a patch?I assume the shQuote change obviates the need for my kludge to texi2dvi. If it were not written in perl I would have looked at the script part too but relearning perl provides too high an obstacle timewise. It seems that the direction is to move the scripts to R anyways so I assume perl will ultimately be eliminated from the R distribution simplifying this sort of maintenance.
I don't know the issue, but a search for texi2dvi under my TeXLive tree found a shell script by that name in the directory texmf\doc\graphics\texdraw. Can that be made to work, in conjunction with the shell in the Rtool bundle? Just curious. (This is fpTeX installed from the TeXLive 2004 CD.) Andy> From: Gabor Grothendieck > > From: Uwe Ligges <ligges@statistik.uni-dortmund.de> > > Gabor Grothendieck wrote: > > > > > I noticed a shQuote fix for Windows in the svn logs. > > > Just wanted to point out that this will favorably > > > affect texi2dvi on Windows which previously used > > > UNIX quoting and so generated an incorrect Windows > > > command. (Note that texi2dvi is used when creating > > > vignettes.) > > > > > > Another problem is that the recommended tex > > > distribution for Windows, fptex, does not have texi2dvi > > > in the first place. The alternative, MiKTeX, is harder > > > to install > > > > I heared some people think so, but apparently it is not a > big problem. > > I installed both so I know from experience that MikTex was > harder to install. > > The problem is that its a lot of work to set up a package > building environment on Windows. There are numerous systems to > install, TeX, UNIX tools, perl, Microsoft Help plus path problems. > > MikTex is just one > more obstacle. MikTex seems like its a slicker distribution > package so its too bad its more work to set up. The main > problems with fptex are not configuration but that (1) the > package is so large that it takes a long time to download and > (2) its missing crucial elements, viz. texi2dvi. > > > > > > > > than fptex but instructions are available > > > and it does have an equivalent to texi2dvi called texify.exe; > > > however, R still does not know about texify.exe > > > resulting in additional installation hassles, > > > viz. setting the texi2dvi option or setting up a texi2dvi.bat > > > file yourself that calls texify.exe or just forgetting > > > about texi2dvi and manually running the necessary tex commands > > > when creating vignettes on Windows. > > > > texi2dvi is also available in MikTeX, see package > miktex-texinfo-bin. > > > > See also ?texi2dvi in R to see how you can set something > different, e.g. > > "texify", which is explicitly given in the examples. > > > > > > > > > There are also additional problems related to using the > > > perl package-building scripts in Windows. I can create the > > > zoo vignette using: > > > > > > setwd("...whatever...") > > > Sweave("zoo.Rnw") > > > texi2dvi("zoo.tex") > > > > > > provided I modify the source to texi2dvi to use the > > > appropriate quoting but even with that modification, > > > the perl package build scripts only build > > > the vignette as far as the zoo.tex file and I must > > > do the rest by hand so there is some problem with > > > the scripts or in how they interact with R. > > > > Are you going to provide a patch? > > I assume the shQuote change obviates the need for my kludge to > texi2dvi. If it were not written in perl I would have looked > at the script part too but relearning perl provides too high > an obstacle timewise. It seems that the direction is to move > the scripts to R anyways so I assume perl will ultimately be > eliminated from the R distribution simplifying this sort of > maintenance. > > ______________________________________________ > R-devel@stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel > >