Hi Michael, You have a good point. Retranslation could be done when the spacebar is pressed or after the user has not typed for a second or so. However, for text nodes of reasonable length retranslating only from the point of the change is probably not justified by the extra complication. I am copying this message to the list, since I think that is where it should be. John On Fri, Jan 04, 2013 at 10:09:53AM +0000, Michael Whapples wrote: > I don't know what my initial view was, but certainly looking at it now, > does this seem sensible? > > What I mean by this, is it wise to retranslate on every character key > press? There will be points where it does not logically make sense to > translate it as it will not stay that way. An example is if I were > typing a word, until the word is complete I am not really bothered about > how it has translated it as I will be adding more characters and the > translation probably will change very soon. > > Going with the average word length being approximately 5 characters (I > know the precise figure may differ but this is close enough and makes > the maths easy), and assuming someone is only typing complete words (not > correcting spelling errors), then approximately 80 percent of > translations will be rubbish and quickly replaced. > > Also in most cases the translation before the insertion point will not > change, this seems like a huge amount of retranslation which is just > unnecessary. > > So I will ask the question, when is it the user wants a translation > done? Once we know that then we can try and find the event which relates > to that want. > > While being a bit more complicated to implement, I would have thought a > more efficient thing would to wait for the user to stop typing before > retranslating (eg. if they stop typing for a second or more). This still > may not quite be when they want the translation done (it is my guess of > when I would want it), so back to the question of when do they want it. > > Michael Whapples > On 03/01/2013 20:03, John J. Boyer wrote: > >We've gone through this before. I thought we had agreed that each text > >node in an xml file would be dynamically retranslated whenever a > >character key was pressed. This might throw page breaks, etc. off, but > >the document could be re-rendered as a whole when it is saved. I've > >worked out a new function for liblouisutdml to do translate single text > >nodes, but not > >implemennted it yet. This would be fast and shouldn't cause a noticeable > >perfourmance hit. > > > >There is no need for radical changes. Just the new function. > > > >John V > > > >On Thu, Jan 03, 2013 at 06:41:04PM +0000, Keith Creasy wrote: > >>Someone explain to me why we have to pump the entire xml document through > >>LibLouisUTDML other than for pagenation and volume breaks? I'm thinking > >>that we don't but maybe I'm wrong. > >> > >>I have some ideas but they aren't fleshed out yet. > >> > >>Keith > >> > >> > >> > >>Keith Creasy > >>Software Developer > >>American Printing House for the Blind > >>KCreasy@xxxxxxx > >>Phone: 502.895.2405 > >>Skype: keith537 > >> > >> > >>-----Original Message----- > >>From: John Gardner [mailto:john.gardner@xxxxxxxxxxxx] > >>Sent: Thursday, January 03, 2013 1:17 PM > >>To: 'Michael Whapples' > >>Cc: Keith Creasy; 'John J. Boyer' > >>Subject: RE: Steering committee > >> > >>Michael, yes there are lots of good reasons that you don't want to send > >>page-by-page info to any printer. As for device-independence, yes it is > >>true that the utdml Braille formatting is done for a particular page > >>size, line length, etc. But if one wants to send it to another device, > >>you just send the file back to liblouistdml with different parameters. > >>At least liblouistdml is fast. > >> > >>I really don't see any advantage to doing anything else if the goal is > >>getting embossed Braille. I just don't know whether it is practical to > >>use liblouistdml for on-the-fly Braille translation/formatting. Probably > >>not because it will spend too much time re-processing previous pages. So > >>what one would need is some kind of differential liblouistdml that would > >>accept everything up to page x and then translate/format everything > >>beyond that point. Maybe not elegant, but it might be better in some > >>ways than using style sheets. Thoughts from JJB and Keith? > >> > >> > >> > >>John > >> > >> > >>-----Original Message----- > >>From: Michael Whapples [mailto:michael.whapples@xxxxxxxxxxxx] > >>Sent: Thursday, January 03, 2013 10:06 AM > >>To: John Gardner > >>Cc: 'Keith Creasy'; 'John J. Boyer' > >>Subject: Re: Steering committee > >> > >>I guess with on the fly formatting, the speed would depend on how much > >>one had to get done. > >> > >>An example would be if the on screen display only showed one Braille page > >>then it only need format one page before it can show it, the rest could > >>be done in the background or if it were fast enough done as those pages > >>come into view. > >> > >>Whether for embossing one could actually start sending the first page to > >>the embosser while formatting the next page, and so on I don't know. I > >>get the feeling for ViewPlus embossers, possibly not as it goes through > >>the Windows print system and for that doesn't one need to send the whole > >>file at once, otherwise there may be a risk that something else may send > >>its document between your pages (IE. if you sent a page at a time by > >>sending a file which contained each page). > >> > >>Although I have to say it does sound a nice idea if one could have a > >>Braille translation of a document which was device independent and it is > >>formatted when one decides what device they will be sending it to. > >> > >>Michael Whapples > >>On 03/01/2013 17:55, John Gardner wrote: > >>>I'm not quite following the subtle details of this discussion Keith. > >>>What I think I hear is that you would like an architecture that would > >>>send text to liblouis (or other) to be translated as a step in the > >>>process. Then as a separate step, you would allow CSS or XSLT to > >>>format that Braille on the fly. I see no role for utdml in that > >>>process. There is nothing wrong in principle with doing it your way, > >>>but there would need to be a big development project to get the > >>>formatting > >>right. > >>>FYI there is a French-led project to translate math using CSS. I > >>>understand that it does a good job on French math and maybe others, > >>>but it is apparently very very slow. And it is not intended to be 'on > >>>the fly'. Is speed gonna be an issue for on-the-fly formatting? > >>> > >>>John > >>> > >>> > >>>-----Original Message----- > >>>From: Keith Creasy [mailto:kcreasy@xxxxxxx] > >>>Sent: Thursday, January 03, 2013 7:46 AM > >>>To: John Gardner; 'Michael Whapples'; 'John J. Boyer' > >>>Subject: RE: Steering committee > >>> > >>>Hi John. > >>> > >>>The disagreement is more about how and where UTDML gets applied I > >>>think. I assume we'll be using LibLouis UTDML perhaps with some > >>modifications. > >>>Yes, I do think translation and formatting are two separate things but > >>>that's a different issue. My intent is to make them separate at some > >>>point even if it means extending BrailleBlaster. > >>> > >>>As far as mainstream applications formatting braille correctly I'm not > >>>sure what you are asking. No application that doesn't have rules for > >>>braille are going to do it right. In that sense I do agree. I would > >>>submit however that CSS and XSLT as well as other technologies are > >>>pretty powerful and could go a long way toward rendering braille > >>>correctly from any XML document type. I even had a discussion with > >>>Brandon this morning about the possibility of using a web-based UI > >>connected to a service. > >>>In any case I think we've decided, once again, that we'll let > >>>BrailleBlaster go where you, Michael, and John B. want it to go and > >>>we'll contribute as we can. If we deccide it doesn't really work well > >>>enough in a production environment then we'll extend it or simply > >>>create another UI for LibLouis and LibLouis UTDML. > >>> > >>> > >>> > >>>Keith Creasy > >>>Software Developer > >>>American Printing House for the Blind > >>>KCreasy@xxxxxxx > >>>Phone: 502.895.2405 > >>>Skype: keith537 > >>> > >>> > >>>-----Original Message----- > >>>From: John Gardner [mailto:john.gardner@xxxxxxxxxxxx] > >>>Sent: Thursday, January 03, 2013 10:28 AM > >>>To: Keith Creasy; 'Michael Whapples'; 'John J. Boyer' > >>>Subject: Steering committee > >>> > >>>Hi, I took this discussion off the list for a short steering committee > >>>discussion. I fully approve of BB as an XML editor - it has been from > >>>the beginning, but also from the beginning we wanted it to look and > >>>feel like a simple WYSIWYG to novices. I think we are all on the same > >>page here. > >>>Where we are not on the same page is about the utdmo format, and I am > >>>not enough of a software person to know how to bridge this gap. Utdml > >>>was devised to be a format for embossed Braille, having all the > >>>Braille formatting embedded. Admittedly it is a strange beast, but > >>>the formatting is 90% of the problem in Braille "translation" > >>>according to my friend Susan Jollly. I agree with Susan in this respect. > >>> > >>>My understanding is that Keith would like to have Braille formatted > >>>separately from translation and to have it done in BrailleBlaster, not > >>>in liblouistdml. As long as we get properly-formatted embossed > >>>Braille out, I guess it wouldn't really matter. But in fact the > >>>liblouistdml routine that does both is working pretty well now, and > >>>ViewPlus is using it in virtually all new developments. It would be a > >>>big problem for us to reverse course unless we had a really big reason > >>>to > >>do it. > >>>So Keith, I guess I put this into your court. I do not believe that > >>>any standard mainstream application can format Braille and get it > >>>right. Am I wrong? > >>> > >>>John G > >>> > >>> > >>>-----Original Message----- > >>>From: brailleblaster-bounce@xxxxxxxxxxxxx > >>>[mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Keith Creasy > >>>Sent: Thursday, January 03, 2013 6:50 AM > >>>To: brailleblaster@xxxxxxxxxxxxx > >>>Subject: [brailleblaster] Re: Different Approaches to document > >>>processing. > >>> > >>>Michael. > >>> > >>>What other design patterns/practices do you think we should consider? > >>> > >>>Yes, I agree 100% that whenever possible we should use established > >>>mainstream design practices and standards, like MVC CSS, and XSLT, > >>>because a lot of developers already know and understand them. > >>> > >>>About XML editing, don't confuse the term XML Editor" with something > >>>that has to be complicated and technical. It can still be fairly > >>>simple. Also, I've always felt that it should be something that > >>>doesn't have to be used by people who just want to write and print > >>>braille. Since we are editing XML > >>>(UTDML) then one could posit that BrailleBlaster is an XML editor > >>>whether it appears that way or not to users. > >>> > >>> > >>>Keith Creasy > >>>Software Developer > >>>American Printing House for the Blind > >>>KCreasy@xxxxxxx > >>>Phone: 502.895.2405 > >>>Skype: keith537 > >>> > >>> > >>>-----Original Message----- > >>>From: brailleblaster-bounce@xxxxxxxxxxxxx > >>>[mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael > >>>Whapples > >>>Sent: Thursday, January 03, 2013 9:36 AM > >>>To: brailleblaster@xxxxxxxxxxxxx > >>>Subject: [brailleblaster] Re: Different Approaches to document > >>>processing. > >>> > >>>Hello, > >>>I have had some concerns over the design of BrailleBlaster for a > >>>little bit, that is partly why I decided to keep away from dealing in > >>>the code too much and restrict myself to managing the project site and > >>repositories. > >>>Keith I know you mentioned about MVC, yes that probably would be a > >>>good idea, but may be BrailleBlaster needs to make use of other design > >>>patterns and other good practice as well. By neglecting these things > >>>its probably made the code and design much harder to follow and means > >>>that where one might use standard terms these might not apply or may > >>>even have a different meaning and so leading to miscommunications. > >>> > >>>Regarding the use of semantic actions, I would dispute the arguement > >>>that they are easier to use than XSL. My reasoning here again goes > >>>along the line of using something familiar to people, XSLT is widely > >>>used and so potential developers may already be familiar with it, > >>>should one have a difficulty in making it do what they need then it is > >>>much more likely they will find an answer (even if the answer is "not > >>>possible") by searching the internet as there is much more information > >>>out > >>there on XSLT. > >>>Keith you mention about defining a document interface, yes this is > >>>something I have felt for some time, the view should have no need to > >>>understand the underlying document format, it should deal with a > >>>single document interface which through various adaptors/providers > >>>will modify the document in the appropriate way. As an example, the > >>>view should only deal with headings, it should not be concerned with > >>>how > >>HTML or DocX store a heading. > >>>My understanding at present is that the semantic actions miss out the > >>>document interface stage, mapping straight from a file format to how > >>>they should be presented. This step really should be split, actions > >>>for adapting the format into the document interface and actions for > >>>how the view will render the document structure element. The former of > >>>these would not need to be customisable by users (only developers who > >>>want to add support for additional formats need work with it), whereas > >>>the latter would be desirable to be customised by users (eg. sighted > >>>users may want headings shown in a different colour, but those who are > >>>colour blind may wish to configure it so that headings can be > >>>distinguished without the use of colour). If one does not separate > >>>these concerns then users may need to deal with every file format they > >>>wish to use, meaning additional work for the user and potentially > >>>requiring them to have fairly advanced knowledge we really should not > >>expect of users. > >>>As for whether BrailleBlaster should have a XML editor, my view would > >>>be not. Such an editor would be contrary to the initial aims of > >>>BrailleBlaster being easy to use by those with little computer > >>>knowledge (IE. not expect more knowledge than knowing how to use a > >>>word processor like Word), as an XML editor would require > >>>understanding of the underlying XML format. As mentioned, if someone > >>>wishes to have such an editor then there are probably other tools more > >>suited for doing that XML editing. > >>>Also I feel by spending time on features not really adding to what > >>>BrailleBlaster is aimed to achieve will just delay BrailleBlaster ever > >>>reaching the goal of what we want it to be. > >>> > >>>I know some may feel I possibly have a restricted view of what > >>>BrailleBlaster should be, my response is that I feel keep it focused > >>>to do a particular task for a particular type of user and do that very > >>>well. If one tries to make something which is everything for everyone, > >>>then it runs the risk of actually not working well for anyone (IE. its > >>>too complicated for beginners and have annoying features for experts). > >>>There is certainly some software I can think of which falls in the > >>>category of trying to be everything to everyone but not working well > >>>for anyone, where competing products while more restricted work far > >>better. > >>>There possibly is more, but I possibly have raised it in the past with > >>>little affect. So in that spirit, I have pretty well said what I feel, > >>>unless further discussion would go anywhere I don't really have much > >>>apitite to get into big discussions now. In general I think Keith is > >>>making some reasonable points. > >>> > >>>Michael Whapples > >>>On 03/01/2013 12:45, Keith Creasy wrote: > >>>>Yes, I've read the comments in the code and your architecture file. > >>>>Very > >>>helpful. > >>>>We need to work on the interface between the views and the document. > >>>>That > >>>probably means defining a document interface and a view interface that > >>>is XML centric. > >>>>John, I'm not sure what anyone else can contribute since you want > >>>>control > >>>of the word processor and the document model. Until those are fleshed > >>>out there's not really much anyone else can do. At least, speaking for > >>>myself, I can't think of anything substantive I can contribute. > >>>>I'm beginning to think that an XML editor really has no place here. > >>>>Anyone > >>>can just use an external XML editor if they need to do that. > >>>> > >>>> > >>>> > >>>>Keith Creasy > >>>>Software Developer > >>>>American Printing House for the Blind KCreasy@xxxxxxx > >>>>Phone: 502.895.2405 > >>>>Skype: keith537 > >>>> > >>>>-----Original Message----- > >>>>From: brailleblaster-bounce@xxxxxxxxxxxxx > >>>>[mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of John J. > >>>>Boyer > >>>>Sent: Thursday, January 03, 2013 1:50 AM > >>>>To: brailleblaster@xxxxxxxxxxxxx > >>>>Subject: [brailleblaster] Different Approaches to document processing. > >>>> > >>>>Hi Keith, > >>>> > >>>>Yes. Please do outoline your approach. There may be some here who are > >>>>not > >>>acvaquaintged with it. My approach is described in detail in the files > >>>in the documentation directory of the brailleblaster.newdesign > >>>repository. In addition, Semantics.java contains considerable > >>>documentation in the form of javadoc comments. > >>>>Thanks, > >>>>John > >>>> > >>>>-- > >>>>John J. Boyer; President, Chief Software Developer Abilitiessoft, Inc. > >>>>http://www.abilitiessoft.com > >>>>Madison, Wisconsin USA > >>>>Developing software for people with disabilities > >>>> > >>>> > >>>> > >>> > >>> > >>> > >>-- > >>________________________________ > >> > >>Michael Whapples | Software engineer | ViewPlus > >>541.754.4002 x 282 | www.viewplus.com > >>________________________________ > >>PRIVILEGED AND CONFIDENTIAL: This message and any files transmitted with > >>it may be proprietary and are intended solely for the use of the > >>individual to whom they are addressed. If you are not the intended > >>recipient, any use, copying, disclosure, dissemination or distribution is > >>strictly prohibited; please notify the sender and delete the message. > >>ViewPlus Technologies, Inc. > >>accepts no liability for damage of any kind resulting from this email. > >> > > -- > ________________________________ > > Michael Whapples | Software engineer | ViewPlus > 541.754.4002 x 282 | www.viewplus.com > ________________________________ > PRIVILEGED AND CONFIDENTIAL: This message and any files transmitted with it > may be proprietary and are intended solely for the use of the individual to > whom they are addressed. If you are not the intended recipient, any use, > copying, disclosure, dissemination or distribution is strictly prohibited; > please notify the sender and delete the message. ViewPlus Technologies, > Inc. accepts no liability for damage of any kind resulting from this email. -- John J. Boyer; President, Chief Software Developer Abilitiessoft, Inc. http://www.abilitiessoft.com Madison, Wisconsin USA Developing software for people with disabilities