mailing-list for TeXmacs Users

[luis jaime salazar ramírez <address@hidden>]

Hola Pablo:

Gracias por tu respuesta, la verdad yo he detectado varios de los
problemas que me transcribes, pero no los había escrito por falta de
tiempo, y aunque no lo quería aceptar, creo que la opción es esta, es
decir, escribir en TeXmacs y refinar en LaTeX.

Como dice el texto, no hay formas sencillas para usuarios noveles para
que manejen el programa en asuntos que ya están estandarizados y hace
falta solucionar problemas de estabilidad del programa.  Lo anterior no
quiere decir que no me parezca una herramienta importante para disminuir
el tiempo de edición:  lo que antes en LaTeX hacía en 4 horas, ahora lo
puedo hacer con TeXmacs en menos de dos horas y eso para mí es
importante, tal vez por ello me he acostumbrado a esquivar las
combinaciones de teclas que hacen que el programa se caiga, y alguno que
otro aspecto indeseable, etc.

Me gustaría saber qué pasó con este hilo, que ha respondido Joris al
respecto.

Gracias, nuevamente por tu ayuda.

att,


Luis Jaime



El lun, 11-07-2011 a las 17:19 +0200, Pablo Angulo escribió:
> Hola!
> 
> Te copio-y-pego un email que llegó a la lista hace unos meses, de un
> tipo que acababa de escribir su tesis con texmacs. Yo estoy en la misma
> tesitura y casi que me inclino por escribir los capítulos por separado y
> pasarlo a latex cuando termine:
> 
> > I just finished a major project (thesis) using TeXmacs, and I wanted
> > to share some experiences. The project has about 200 pages and 40
> > floats. I have been using TeXmacs for several years now, and until now
> > regularly used TeXmacs for the "draft" phase of various documents (for
> > which it really shines), and used LaTeX for the final typesetting.
> > However, this time I chose to use TeXmacs from the beginning to the
> > end. I found myself developing so many workaround that I thought I'd
> > share them, in case anyone wants to take a similar path, and so that
> > maybe Joris (or someone else who knows the sources) can fix some of
> > the bugs.
> >
> > Don't get me wrong, I like TeXmacs, and it works very well for small
> > documents. However, I think TeXmacs can only compete with LaTeX if it
> > provides a solid stable foundation that users and publishers can
> > depend on. While LaTeX needs a huge amount of work in order to get an
> > acceptable workflow, when correctly set up it is rock solid. TeXmacs
> > is easy to get used to, but extremely crashy when trying to do
> > non-basic operations. With hindsight, I have to say that had I known
> > all the problems I have encountered, I had probably chosen LaTeX, from
> > which I know that it is hard to get used to, but it is stable. It is
> > very frustrating having to deal with mysterious crashes and figuring
> > out workarounds close to the deadline.
> >
> > Currently it seems that new features are added all the time, such as
> > the vector graphics editor. In my opinion TeXmacs has more than enough
> > innovative features and flexibility by now to beat any other
> > scientific editor, but long-term usability is just bad. With respect
> > to the list below, I think that the upmost priority should be to get
> > the existing code as stable and fast as possible (in particular for
> > large documents), and to get the stylesheet language as solid and as
> > well-documented as possible. Otherwise TeXmacs will never get a
> > substantial user base.
> >
> > That said, this is the list. It is organized in three parts: dangerous
> > things than can ruin the output by silently introducing errors,
> > annoying things that are just counter-intuitive or take a lot of time
> > getting used to, and performance issues that are relevant when editing
> > large documents. I loosely ordered the items in order of descending
> > importance.
> >
> > 1) Dangerous things
> > -------------------
> >
> > 1a) Deleted references: TeXmacs indefinitely remembers references it
> > has once seen. If I create a <label>, reference it using <reference>
> > and then delete the <label>, the reference is unchanged (even after
> > Document > Update > All, multiple times), while it should display "?"
> > instead. This can create very hard-to-spot errors in your document. As
> > a workaround, I manually opened the .tm file and deleted everything in
> > the <\references> and <\auxiliary> sections, which forces TeXmacs to
> > regenerate everything, and searched for "?" in the PDF to check if
> > there are any undefined reference. However, this should never happen
> > in the first place. Moreover, there should be an indicator/warning if
> > there are any undefined references or "cite"s in the document.
> >
> > 1b) .tm~ Files: By default, the "File > Load" (BTW, why not "Open"?)
> > dialog shows all file types (in fact there is no option to restrict it
> > to TeXmacs files, e.g. .tm + .ts). In particular, it shows the .tm
> > files side by side with their .tm~ files. It is very easy to mis-click
> > the .tm~ file, which in an unfortunate setting may mean that you lose
> > all of your work. TeXmacs should either not allow to open .tm~ files
> > at all, hide them by default, or at least issue a warning when the
> > user tries to open one.
> >
> > 1c) Page Setup: There are two places where the page size can be set:
> > Under File > Page Setup and under Document > Page > Size. At the
> > least, the menu should be called "Print setup". Such things sound like
> > minor issues, but anyone will FIRST find the prominent "Page setup"
> > menu, change the size there, and assume that it did the trick, when
> > all that is changed is the print setup. This issue alone cost me 15
> > hours of re-breaking the whole document close to the deadline.
> >
> > 1d) Document > Magnification: This is a similarly dangerous item,
> > which should at least issue a warning that it changes the FONT SIZE,
> > i.e. the amount of text that fits on a page. This is very easy to
> > confuse with View > Shrinking factor. If you are required to submit an
> > 11pt document and accidentally mess with the "Magnification" you are
> > in trouble, since TeXmacs will display 11pt font size, but it won't be
> > 11pt in print.
> >
> > 1e) Consistency: I do not know how exactly, but I managed to get two
> > installations of TeXmacs 1.0.7.10 that work exactly the same, except
> > that in one copy the *bold* fonts (and only those) are a tiny bit
> > wider. Therefore headings (and figures for captions) wrap differently.
> > I have no idea if the font was changed recently or if the issue is
> > caused by the setup, but in order to be exchangeable TeXmacs should
> > make absolutely sure that documents look 100% the same when compiled
> > on different systems (maybe use a checksum on the font files?). This
> > one took me one day to figure out.
> >
> > 1f) Sums & Integrals: There seems to have been a semantic change with
> > respect to sums and integrals, which are now considered as "functions"
> > (i.e. they have an argument), rather than just a symbol. This
> > introduced a nasty behavior: If one creates an integral (say
> > <big-around|<int>|<rsub|0><rsub|1>xyz>), puts the cursor at the end
> > and presses "backspace" in order to delete the x, the <int> sign at
> > the front is deleted instead! This is actually very easy to miss if
> > you press "backspace" multiple times, e.g. in order to delete multiple
> > characters, and corrupts the formulas. Instead, the last character
> > should be deleted.
> >
> > 1g) PDF Export: Embedded raster images (especially black & white) look
> > ugly in the exported PDF. They seem to be compressed as very
> > low-quality JPG, with clearly visible JPEG artifacts especially in
> > purely black/purely white regions. The only workaround is to export as
> > PS and convert to PDF manually using Ghostscript:
> >
> > gswin32c -sPAPERSIZE=a4 -sDEVICE=pdfwrite -q -dBATCH -dNOPAUSE -dSAFER
> > -dPDFX -dPDFSETTINGS=/prepress -dAutoFilterColorImages=false
> > -dColorImageFilter=/FlateEncode -dAutoFilterGrayImages=false
> > -dGrayImageFilter=/FlateEncode -sOutputFile=thesis.pdf -c save pop -f
> > thesis.ps
> >
> > There should be a menu item to change the "PS to PDF" command, and a
> > better preset.
> >
> > 1h) Document > Update: One of the things that annoy me most when using
> > LaTeX is that it doesn't automatically
> > perform another run if it is required in order to get all the
> > references right. However, LaTeX at least issues a warning. Using
> > TeXmacs there is no warning, but the same annoying process of Document
> > * click * Update * click * all * wait * Document * click * Update *
> > click * etc. about 5 times in a row to be sure that all references are
> > up-to-date. In my opinion, TeXmacs should automatically perform as
> > many runs as required so that the labels do not change anymore.
> >
> > 1i) Breaking of inline math: Inline formulas that are wrapped around a
> > line ending are just plain ugly. For this reason LaTeX does not wrap
> > them by default, and issues an "overfull hbox" warning instead.
> > TeXmacs wraps them quietly, which requires to manually check every
> > single line if it contains a silently wrapped inline formula.
> >
> > 1k) Style files: I use a custom macro package for shortcuts, pretty
> > formatting of algorithms, etc. All documents are in a Subversion
> > repository, so it regularly happens (by an "svn update") that a style
> > file changes without TeXmacs noticing (TeXmacs is not running during
> > the "svn update"). I regularly found TeXmacs marking "undefined
> > macros" in the main document, when they were cleary defined in the
> > package file. It took me about 2 months to figure out that the styles
> > need to be refreshed using "Tools > Update > Styles", and that
> > restarting TeXmacs does NOT help. Why this doesn't happen
> > automatically when TeXmacs is restarted is completely
> > counter-intuitive. However, it may mess up your whole document, or
> > make it use old macros, which both is very dangerous.
> >
> > 2) Annoying things
> > ------------------
> >
> > 2a) Stylesheet Language: I had to implement some extensions using the
> > stylesheet language. If one goes one step beyond the standard
> > functionality, TeXmacs crashes at the blink of an eye. I found myself
> > spending about 20% of the time figuring out a solution for a problem,
> > and then spending 80% of the time trying 4 other solutions until I
> > find one that doesn't crash TeXmacs. While LaTeX can also be
> > incredibly tiresome to work with, it is solid as a rock. I think if
> > TeXmacs aims at production quality and use in professional
> > environment, these issues -- which mostly concern power-users --
> > desperately need to get fixed. Also, there needs to be a clear
> > documentation of the evaluation order of the document tree. I spent
> > several hours trying to figure out the exact semantics of the various
> > "quote" tags, and eventually gave up.
> >
> > 2b) Images: I found that the only image type that can be reliably used
> > in Texmcas is EPS written by Ghostscript. PNG tends to crash TeXmacs
> > with a memory overflow when exporting to PS, I found myself finally
> > converting all raster images to EPS in order to get it to work. EPS
> > files written by other applications (e.g. Illustrator) often result in
> > zero-byte .PS files, or .PS files that are truncated at the page that
> > contains the EPS. The workaround is to run Ghostscript's "eps2eps" on
> > all .EPS files, which creates .EPS files that TeXmacs doesn't crash on.
> >
> > 2c) Menu Layout: Generally, I found that TeXmacs menus are laid out in
> > the way TeXmacs thinks about the document, not in the way that is most
> > intuitive for the user. For example, inserting a float is done via
> > "Format > Page > Insertion > Floating Object". I think the "Format"
> > menu is the last place that anyone would look, in particular when
> > there is a top-level "Insert" menu. Other examples: Tools > Update
> > menu and Document > Update menu with similar functions but at
> > different places, the same holds for Edit > Preferences > Keyboard and
> > Edit > Preferences > Mathematics > Keyboard.
> >
> > 2d) Bugs when using parts: There is an ugly bug when working with
> > parts that causes TeXmacs to reproducibly crash at harmless actions
> > such as selecting text across a chapter boundary (see my previous
> > posts on this list).
> >
> > 2e) Yes/No Dialog: Is there any reason for the cumbersome "yes/no"
> > prompt? Having to backspace away the "no" and type "yes" is very VERY
> > annoying. Even a simple "y"/"n" choice with one keypress would have
> > been more usable. Similarly for the strange Qt "yes/no" dialog with
> > the combobox. This should just use the default yes/no dialog that is
> > provided by all OS and by Qt.
> >
> > 2f) Extensible brackets: With the automatic extensible brackets there
> > was apparantly introduced a bug: When extensible brackets are turned
> > off (Edit > Preferences > Keyboard > Mathematics > Use extensible
> > brackets), I am unable to insert extensible brackets anymore, even
> > using the toolbox icon ("Insert large delimiter"). On the other hand,
> > if I turn extensible brackets on, I cannot type left angles anymore
> > ("<" + 5x TAB), it just displays as "<left-<langle>-0>". I ended up
> > copy & pasting either the large brackets or the angles each time I
> > needed them.
> >
> > 2g) Editing large documents: TeXmacs should be able to view separate
> > parts of a document in different windows. Currently this is not
> > possible, which makes creating references in large documents very
> > slow: Switch to other part * wait * look up label * Switch back to
> > current part * wait * insert label. The same holds for copy & pasting.
> > Editing the whole document in one window is too slow.
> >
> > 2h) Search: The search function needs a window, which should be
> > non-modal. A very common task is to search for all occurrences of a
> > text and change something close to that text which cannot be automated
> > (for example search for all "TODO" and remove them, fix them, or
> > ignore them). The problem is that after typing something, the
> > (apparently undocumented) "Search next" shortcut (F3) does not work
> > anymore, and one has to type Ctrl+F + the whole search phrase again,
> > every single time. The layout for the "Search" dialog box is be quite
> > standard by now (at least on Windows systems): "Match Case",
> > "Up/Down", "Find Next", "Cancel". The same holds for the "Search &
> > Replace" function.
> >
> > 2i) Windows Version: I couldn't get the Windows version to compile, I
> > ended up installing TeXmacs under Ubuntu in a VirtualBox VM, and using
> > a shared folder to access the .tm files on the Windows drive.
> >
> > 2k) Redraw errors: When I show the whole document (Part > Show all
> > parts), I frequently get redraw errors, where text from the current
> > page and text from a completely different section at the end of the
> > document are draw on top of each other, resulting in garbage text. The
> > workaround is to force a redraw e.g. using "page down" + "page up".
> >
> > 2l) Zooming: Zooming in and out should be seamless e.g. using
> > Ctrl+scroll wheel, as is standard in most applications, instead of the
> > "View > Shrinking factor" workaround.
> >
> > 2m) TeXmacs frequently loses its configuration and displays the
> > "Welcome to your new TeXmacs system" screen.
> >
> > 3) Performance
> > --------------
> >
> > 3a) Editing anything above about 15 pages in "Paper" mode is just too
> > slow to be usable. A part of the problem is that TeXmacs updates the
> > whole document at every keypress. Since this cannot always be avoided,
> > e.g. when re-wording some sentences in order to get a prettier line
> > wrap, I ended up typing text fragments in an external editor, and copy
> > & pasting them into TeXmacs as a block in order to avoid the update at
> > every keypress. If views the whole document ("Parts > All") and
> > accidentally types a complete word (with, say 10 characters), it is
> > faster to kill TeXmacs, restart it and reload the document than having
> > to wait for the updates.
> >
> > 3b) Editing paragraphs that contain floats or captions of figures that
> > contain images is incredibly slow, even in very small documents.
> >
> > 3c) The same holds for the "Papyrus" mode when some floats are
> > present. The unsatisfying workaround is to insert only figures while
> > working with the text, and converting them to floats just before
> > creating the final PDF.
> >
>