-----Original Message----- From: Davide P. Cervone [mailto:[EMAIL PROTECTED] Sent: Monday, December 19, 2005 7:29 PM To: Bill Page Subject: Re: [Axiom-developer] RE: tex4ht and jsmath
Bill: Thanks for CCing me on the conversation about jsMath in the Axiom- Developer mailing list. I read through the recent postings, and I might be able to shed some light on the subject. Could you pass these comments on to the list? I'd appreciate it. There are several things involved with the font issues that some of you are having. One of them is the font change that occurred with the v3.0 release of jsMath. This release has new fonts that use a different encoding from the original BaKoMa fonts, and so I have named them with a "jsMath-" prefix to differentiate them, and avoid name collisions with the original ones. Older versions of jsMath will not recognize or use these new fonts, so if you have downloaded fonts starting with "jsMath-", you will still get the font warning message if you view the jsMathExperiment site or http://page.axiom- developer.org/jstest.html (which use jsMath v1.8, which is now quite old), but will NOT get a font warning message on the jsMath examples page (which uses jsMath v3.0). Those who reported bad rendering in MSIE are probably suffering from the older version of jsMath; v2.0 is where the image-based fallback was introduced. With 1.8 if the TeX fonts were not available, jsMath would use unicode characters, and these were not entirely successful with things like roots and other delicate spacing (since jsMath doesn't know what font they will be taken from, so can't really know how they should line up). If you want to test things with jsMath versions 2.x or 1.x, you need to get the BaKoMa versions of the fonts, still available on the jsMath font page (http://www.math.union.edu/locate/jsMath/download/ fonts.html). You can install both sets of fonts if you want (the original BaKoMa fonts and the newer jsMath-specific ones). Someone mentioned that he had not installed any fonts but still didn't get the font warning. That is probably because the BaKoMa fonts are the ones that Firefox uses as part of their font suite for MathML. If you installed the fonts for MathML, then jsMath will use them without your needing to install anything else. Someone else talked about MikTeX installations, but these are not the fonts that jsMath uses. It does not use the .mf or .pk files from a TeX installation, but rather .ttf fonts, which are not usually part of TeX itself. (Some previewers may use these, but not a native TeX installation.) In general, jsMath will not have any connection to a native TeX installation. There was some concern over the method of testing for the TeX fonts. I admit that it is a hack, in the sense that it is not a guaranteed method of detecting the presence of the fonts, but in practice it has worked remarkably well. The CMEX10 font has unusually large depth, and this is so unusual that it is pretty indicative of having the right font. So far there has only been one complaint that I have received about not properly recognizing the fonts, and that was very early on, and the test has been adjusted since then. In fact, it has worked well enough that similar test can be used to detect the presence of other fonts (like those in the extra-fonts area of the jsMath download site). So I'm pretty satisfied with that method of detecting the fonts. The problems that you all are facing are not due to that, but to the change in the fonts used by jsMath. Sorry about the complication that has introduced. A second issue is with the jsMath installations on your servers. I am not able to access the ohio-state site, but the conversation indicates that someone at least had tried to make a page that sourced the copy of jsMath.js from the jsMath home page, and they got warning messages about not taking jsMath from the local server. This is not an idle warning, and is not because I didn't want you to be able to do that; it is a restriction based on the browsers' javascript security model, and it will prevent jsMath from working properly. In particular, jsMath consists of more than one javascript file (not just jsMath.js), and jsMath loads additional components as it needs them. For example, the image fallback mode is not downloaded unless you need it. But the method used to obtain these files relies on the other files coming from the same server as the main page (otherwise the browser won't let jsMath look at the files since they come from a different security domain). The upshot is, if you load jsMath.js from a different server, it will be crippled and won't be able to function fully. If you have the TeX fonts installed, jsMath will run OK in this situation (but the user will not be able to open the control panel, for example), however, if you DON'T have the TeX fonts installed, jsMath will not be able to download the code for the fallback method, and so you will not have your mathematics rendered properly (or at all). This may account for why some of you can see it OK and some of you can't: those WITH the TeX fonts would see properly rendered math, and those without would see the TeX source code unrendered. The same thing goes for those who downloaded one of the jsMath example files and tried to run it locally. First, in those examples, the SCRIPT tag that calls in jsMath has a relative URL not an absolute one, so you would need to change that to point to a local copy of jsMath. Second, if you pointed this to the jsMath.js on my server, you will run into the problems I mentioned above of taking jsMath from a non-local server. Some browsers are less secure than others, and some will let you do this, but not all. Finally, you can't just copy jsMath.js to the local machine, but must get a complete installation of jsMath, otherwise files that jsMath needs will be missing, and things like the fallback method for when the fonts are unavailable will not work. Unfortunately, the lack of TeX fonts and the incomplete jsMath installation have combined to make matters worse for you. If you download the complete jsMath distribution, you should be able to run a local copy just fine. Finally, a note about the image mode for jsMath. This does not use a single image of the complete equation (as is the case with most image- based mathematics). Rather it uses separate images of the individual characters in the TeX fonts and creates the typeset mathematics from those just as though they were a real font. So even when jsMath is using its image fallback mode, you will still get the progress message for the math being typeset. I hope you are able to get this working for you. Let me know if you have additional questions. Davide PS, it is possible to extend jsMath to include additional LaTeX environments that you may need. The changes in v3.0 will make this sort of thing easier to do. If Axiom uses a limited number of environments that are not currently supported by jsMath, I might be able to add them in for you. _______________________________________________ Axiom-developer mailing list Axiom-developer@nongnu.org http://lists.nongnu.org/mailman/listinfo/axiom-developer
