- Dillo-dev - lists.mailman3.com

the Http_must_use_proxy change in CVS

by corvid＠lavabit.com

Is something missing? It gives me a warning because we're sending it a char* now.

Nov. 11, 2008

Re: patch: middle-click submits in new window/tab

by corvid＠lavabit.com

Jorge wrote: > Corvid: please explain me a bit about: > > * The patch changes from "Activate" to "Clicked" signal. Partly because it was a click that's passing button stuff, and partly because I may want to experiment with sending right-click through someday for a form menu. > * It also seems to change the logic in DilloHtmlInput::activate() > - it makes me wonder whether pressing enter on other form widgets > may submit. text/password/isindex can, but textarea won't (no connectActivate). > + fix gcc error and re-submit a patch. I guess I'll go back and make a little thing for the button state to live in as well.

Nov. 10, 2008

Re: patch: middle-click submits in new window/tab

by jcid＠dillo.org

On Mon, Nov 10, 2008 at 07:01:21PM +0100, Johannes Hofmann wrote: > On Mon, Nov 10, 2008 at 02:36:48PM -0300, Jorge Arellano Cid wrote: > > On Mon, Oct 20, 2008 at 02:49:41PM +0000, corvid wrote: > > > I could say that it was for the sake of consistency with link behavior, > > > but really it was because it seemed like a dilloishly simple way > > > to get the equivalent of pages-remember-form-data. > > > > The consistency part is a strong argument. > > > > Also interesting is the way we can have a workaround to > > remember form data, before a more usual way of handling it is > > implemented. > > > > > > > > (But: who will ever discover that it exists or recognize what it > > > makes possible?) > > > > It may be stated in the FAQ. > > > > > > > 1) fltkui: it might be better to stuff all of the button event stuff > > > into a bag like fltkviewbase/layout do with EventButton. It certainly > > > looks nicer having something like a shift mask hidden away in a nice > > > tidy box, but I wasn't sure that it was actually worth doing. > > > > Hmm, I'd say if it's going to be a definitive patch, it's > > worth the effort. > > > > > 2) form: carrying around buttonNo+shifted for a long time > > > is not pleasing, but oh well. > > > > Agreed. > > > > > > Now that Johannes is interested in implementing FORM "memory", > > I'd say it can be stored in an URL/form-data array. I haven't > > checked how other browsers (e.g. FF) solve the problem of privacy > > (aka. when it's desirable to remember data), but most probably a > > quick look will shed some light. > > > > BTW, I can't compile the patch here because: > > > > make[2]: Entering directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' > > form.cc: In member function 'void form::Form::addButtonResource(const char*, > > dw::core::ui::ButtonResource*, const char*)': > > form.cc:241: error: cannot allocate an object of abstract type > > 'form::Form::FormClickedReceiver' > > form.hh:124: note: because the following virtual functions are pure within > > 'form::Form::FormClickedReceiver': > > ../dw/ui.hh:338: note: virtual void > > dw::core::ui::ButtonResource::ClickedReceiver::clicked(dw::core::ui::ButtonReso > > urce*, int, int) > > make[2]: *** [form.o] Error 1 > > make[2]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' > > make[1]: *** [all-recursive] Error 1 > > make[1]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur' > > > > > > > > BTW2: > > > > * The patch changes from "Activate" to "Clicked" signal. > > * It also seems to change the logic in DilloHtmlInput::activate() > > - it makes me wonder whether pressing enter on other form widgets > > may submit. > > > > Anyway, it'd be good if you can coordinate with Johannes and decide what > > approach to take, and send a new patch for it. I'd prefer Johannes to keep > > working in the CSS front, and you to make the definitive patch, but this > > is for you guys to decide. > > I won't start on any new stuff (other than CSS) in the near future. > Currently I'm trying to convert the table code to the new style > handling which is a bit hairy but there are no major obstacles yet. > > And I really like the middle-click to submit thing. So I don't have > a lot of motivation to implement real form data remembering anyway. Great! Corvid: please explain me a bit about: * The patch changes from "Activate" to "Clicked" signal. * It also seems to change the logic in DilloHtmlInput::activate() - it makes me wonder whether pressing enter on other form widgets may submit. + fix gcc error and re-submit a patch. -- Cheers Jorge.-

Nov. 10, 2008

Greetings! -- With Patches :-)

by list-jcr＠designtools.org

On Mon, 10 Nov 2008 15:08:43 -0300 Jorge Arellano Cid <jcid(a)dillo.org> wrote: > > [...] > > The first is to fix an over-sight in the following post/patch > > posted to this mailing list: > > http://article.gmane.org/gmane.comp.web.dillo.devel/6248 > > > > The above patch does not take into consideration font size, so > > the result can be microscopic/unreadable fonts. I've added the > > code to scale the font in the same way as you scale other fonts > > in dillo. Though I've read enough of the dillo docs and code to > > grasp the basics of formatting, I'm unsure if my formatting of > > the continuation (split line) is "correct" according to your > > standards. > > > > There are other places in the dillo UI where font handling is > > still incorrect and can result in unreadable fonts. Some of this > > is the fault of fltk, since as far as I know there is no ~/.gtkrc > > equivalent to set defaults for the tool kit. With luck, I'll find > > and fix them. > > This raises an interesting issue. > > It may be necessary to add "ui_font_factor" to dillorc. > > With the patch "as is" things get broken here. The popup menus > get incomplete titles. > By this I assume you mean the text with the new/larger font size in the title/label on the popup menus does not fit within the provided space? I saw this behavior as well. The fact things work without the patch is really a matter of luck, and not a matter of correctness. The correct way to deal with it is to always calculate the size of the text string (with the specified font), and adjust the size of the label/title accordingly. I remember reading in the fltk docs about a member function to calculate X,Y size but since dillo is my first adventure into fltk, I simply cannot recall the function name. --It's one of those "I know I saw it... but where the heck did I see it?" type of things. ;-) > Looking at the result, I realized my personal preference is to > have smaller fonts in the UI, than in rendered pages. This led me > to the "ui_font_factor" solution. > > CVS has now a similar patch that doesn't scale the font > (submitted by corvid some time ago). > > It'd be good to have a patch that takes all the UI elements > into account. Unfortunately, a patch that scales UI fonts may > also need to scale the UI panel and this could take some effort. > > Scaling the UI font without scaling the UI panel may be easier, > and also serve the purpose. > Keeping the UI usable/readable across a multitude of systems and configurations is a different problem than keeping the displayed web pages usable/readable. It's best to address each problem individually, rather than lumping them together. Though it would be a serious change, and require a lot of patching, I think the best solution is to have a two pronged approach: 1.) Font Factor (scaling) 2.) Font Size If only Font_Size is stated, it would be used verbatim. If only Font_Factor is stated, it would scale the font. if both were stated, Font_Factor would take precedence. This would give you prefs for: vw_font_name vw_font_size vw_font_factor ui_font_name ui_font_size ui_font_factor Ummm... with all due respect, I think the currently used "fontname" in the prefs is against the coding conventions stated on the dillo site. Of course, it really doesn't matter to a compiler, but if I'm going to dig into augmenting this code, I need to know the "right" (TM)(C)(R) way to fix it. :-) > > > Unfortunately, I screwed up pasting my > > email address in the "submitter" section due to the fact that > > dillo doesn't actually copy selected text in a form (i.e. with > > the select/paste of the plain old X Window System). If you want > > to correct the submitter on the bug to my email address, it would > > be appreciated. There was no way I could correct it myself. > > > > On all popup menus in dillo, you have a non-typical and > > non-functional "label" (or better said "title") on the displayed > > popup menu. This was my first adventute into fltk, but according > > to the fltk2 docs, if you set the "label" of the popup menu to > > NULL, the "label" will not be shown. You can do it one of two > > ways, (1) setting the lable to NULL in the instantiation of the > > class or (2) using the ->label() member function to set the label > > to NULL. I opted for the former because it's more clean than the > > latter. > > I like those menu title labels in place. They even help me! > They also help newcomers to grasp Dillo's UI. > > Does anybody else feel like those labels are better out? > (if YES, a preference may be added) > Having an option might work, but I'm typically against having knobs just for the sake of having knobs. Keeping things simple and usable is far more important, and this includes keeping the code simple. I tend to follow the path of least surprise. :-) Since I cannot think of another application which has a label/title on Right-Click popup menus, I think having them would surprise new users more than not having them. By having label/title bars on menus, you're treating them like windows, and hence, the behavior is non-typical, inconsistent and unexpected. As for *my* personal opinion, I hate label/title bars since (to me) they serve no purpose. I'm actually running the cwm window manager so even windows don't waste space on unnecessary, non-functional title bars. > > Though there were "Accelerator" key bindings in the labels > > (such as "&PAGE OPTIONS" to get enable win32 meun-ish "<ALT>-P" > > activation), they don't actually work since the popup menu type > > ("POPUP123") means they are completely hidden until activated by > > the mouse buttons. > > There's a FLTK STR on this. > Bummer! It would be sweet if we could activate the popups throught he keyboard. Hopefully this will get fixed. > > There is still a popup menu placement issue (annoyance) with > > the Bug Meter when running dillo in full screen (or to the bottom > > edge of your screen) but it seems to be caused by the fltk > > "remember last selected" option. It does not properly realize the > > bottom edge of the screen/window, so some popup menu options are > > off-screen and you need a mouse-wheel movement to access them. > > This may be a bug in FLTK, but I'm not sure. > > BTW, it's good to see new developers around. Please check a bit > the latest activity in dillo-dev and you'll find other FLTK/UI > issues waiting for a patch. Thanks. I'll try to read the archives a bit more. -- J.C. Roberts

Nov. 10, 2008

Greetings! -- With Patches :-)

by corvid＠lavabit.com

Jorge wrote: > On Sun, Nov 09, 2008 at 12:16:42PM -0800, J.C. Roberts wrote: > > On all popup menus in dillo, you have a non-typical and > > non-functional "label" (or better said "title") on the displayed > > popup menu. This was my first adventute into fltk, but according > > to the fltk2 docs, if you set the "label" of the popup menu to > > NULL, the "label" will not be shown. You can do it one of two > > ways, (1) setting the lable to NULL in the instantiation of the > > class or (2) using the ->label() member function to set the label > > to NULL. I opted for the former because it's more clean than the > > latter. > > I like those menu title labels in place. They even help me! > They also help newcomers to grasp Dillo's UI. > > Does anybody else feel like those labels are better out? > (if YES, a preference may be added) I think I like having the labels.

Nov. 10, 2008

Greetings! -- With Patches :-)

by corvid＠lavabit.com

Jon wrote: > On Mon, 10 Nov 2008 04:59:50 +0000 "corvid" wrote: > > > > With GTK, you can control many aspects of the toolkit such as font > > > names/sizes on all applications through the system-wide or > > > user-level .gtkrc files. > > > > I wonder whether this is why dillo has historically used > > prefs.font_factor instead of a prefs.fontsize (where FontSizes in > > html.cc would be generated from it). > > This is actually boils down to a "personal preference" issue. Some > people believe it's a good thing for a web page to have fonts of > multiple sizes. Being that I can be somewhat difficult, I disagree > strongly, and think all text should be rendered in the exact same font > size... --But of course, that's *my* "personal preference." :-) > > For example, header tags in HTML (i.e. H1, H2, H3, ...) typically use a > larger font. If the dillo developers used prefs.fontsize, the result > would be what *I* like, namely all fonts would be forced to use the > exact same size. > > By using prefs.font_factor, the dillo developers made a wise choice > because it both prevents web developers from using microscopic font > sizes but still allows for variations in font size within a web page. Right, what I meant here is prefs.fontsize as the default as well as umm...h4, iirc, and then figuring out what h1-3,5-6 should be based on that value.

Nov. 10, 2008

Greetings! -- With Patches :-)

by jcid＠dillo.org

On Sun, Nov 09, 2008 at 12:16:42PM -0800, J.C. Roberts wrote: > Hi list, > > I guess the best way to show up on a development list is with > patches, so I've attached two of them. The diffs were created > against cvs with: Good! Welcome aboard. > [...] > The first is to fix an over-sight in the following post/patch > posted to this mailing list: > http://article.gmane.org/gmane.comp.web.dillo.devel/6248 > > The above patch does not take into consideration font size, so > the result can be microscopic/unreadable fonts. I've added the > code to scale the font in the same way as you scale other fonts > in dillo. Though I've read enough of the dillo docs and code to > grasp the basics of formatting, I'm unsure if my formatting of > the continuation (split line) is "correct" according to your > standards. > > There are other places in the dillo UI where font handling is > still incorrect and can result in unreadable fonts. Some of this > is the fault of fltk, since as far as I know there is no ~/.gtkrc > equivalent to set defaults for the tool kit. With luck, I'll find > and fix them. This raises an interesting issue. It may be necessary to add "ui_font_factor" to dillorc. With the patch "as is" things get broken here. The popup menus get incomplete titles. Looking at the result, I realized my personal preference is to have smaller fonts in the UI, than in rendered pages. This led me to the "ui_font_factor" solution. CVS has now a similar patch that doesn't scale the font (submitted by corvid some time ago). It'd be good to have a patch that takes all the UI elements into account. Unfortunately, a patch that scales UI fonts may also need to scale the UI panel and this could take some effort. Scaling the UI font without scaling the UI panel may be easier, and also serve the purpose. > The second patch is for BUG #832 --If I read things correctly, > I'm supposed to file a bug report first, assign it to myself, and > then send in patches. Yeah, but we haven't been strict on that. > Unfortunately, I screwed up pasting my > email address in the "submitter" section due to the fact that > dillo doesn't actually copy selected text in a form (i.e. with > the select/paste of the plain old X Window System). If you want > to correct the submitter on the bug to my email address, it would > be appreciated. There was no way I could correct it myself. > > On all popup menus in dillo, you have a non-typical and > non-functional "label" (or better said "title") on the displayed > popup menu. This was my first adventute into fltk, but according > to the fltk2 docs, if you set the "label" of the popup menu to > NULL, the "label" will not be shown. You can do it one of two > ways, (1) setting the lable to NULL in the instantiation of the > class or (2) using the ->label() member function to set the label > to NULL. I opted for the former because it's more clean than the > latter. I like those menu title labels in place. They even help me! They also help newcomers to grasp Dillo's UI. Does anybody else feel like those labels are better out? (if YES, a preference may be added) > Though there were "Accelerator" key bindings in the labels > (such as "&PAGE OPTIONS" to get enable win32 meun-ish "<ALT>-P" > activation), they don't actually work since the popup menu type > ("POPUP123") means they are completely hidden until activated by > the mouse buttons. There's a FLTK STR on this. > There is still a popup menu placement issue (annoyance) with > the Bug Meter when running dillo in full screen (or to the bottom > edge of your screen) but it seems to be caused by the fltk > "remember last selected" option. It does not properly realize the > bottom edge of the screen/window, so some popup menu options are > off-screen and you need a mouse-wheel movement to access them. This may be a bug in FLTK, but I'm not sure. BTW, it's good to see new developers around. Please check a bit the latest activity in dillo-dev and you'll find other FLTK/UI issues waiting for a patch. -- Cheers Jorge.-

Nov. 10, 2008

Re: patch: middle-click submits in new window/tab

by Johannes.Hofmann＠gmx.de

On Mon, Nov 10, 2008 at 02:36:48PM -0300, Jorge Arellano Cid wrote: > On Mon, Oct 20, 2008 at 02:49:41PM +0000, corvid wrote: > > I could say that it was for the sake of consistency with link behavior, > > but really it was because it seemed like a dilloishly simple way > > to get the equivalent of pages-remember-form-data. > > The consistency part is a strong argument. > > Also interesting is the way we can have a workaround to > remember form data, before a more usual way of handling it is > implemented. > > > > > (But: who will ever discover that it exists or recognize what it > > makes possible?) > > It may be stated in the FAQ. > > > > 1) fltkui: it might be better to stuff all of the button event stuff > > into a bag like fltkviewbase/layout do with EventButton. It certainly > > looks nicer having something like a shift mask hidden away in a nice > > tidy box, but I wasn't sure that it was actually worth doing. > > Hmm, I'd say if it's going to be a definitive patch, it's > worth the effort. > > > 2) form: carrying around buttonNo+shifted for a long time > > is not pleasing, but oh well. > > Agreed. > > > Now that Johannes is interested in implementing FORM "memory", > I'd say it can be stored in an URL/form-data array. I haven't > checked how other browsers (e.g. FF) solve the problem of privacy > (aka. when it's desirable to remember data), but most probably a > quick look will shed some light. > > BTW, I can't compile the patch here because: > > make[2]: Entering directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' > form.cc: In member function 'void form::Form::addButtonResource(const char*, > dw::core::ui::ButtonResource*, const char*)': > form.cc:241: error: cannot allocate an object of abstract type > 'form::Form::FormClickedReceiver' > form.hh:124: note: because the following virtual functions are pure within > 'form::Form::FormClickedReceiver': > ../dw/ui.hh:338: note: virtual void > dw::core::ui::ButtonResource::ClickedReceiver::clicked(dw::core::ui::ButtonReso > urce*, int, int) > make[2]: *** [form.o] Error 1 > make[2]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' > make[1]: *** [all-recursive] Error 1 > make[1]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur' > > > > BTW2: > > * The patch changes from "Activate" to "Clicked" signal. > * It also seems to change the logic in DilloHtmlInput::activate() > - it makes me wonder whether pressing enter on other form widgets > may submit. > > Anyway, it'd be good if you can coordinate with Johannes and decide what > approach to take, and send a new patch for it. I'd prefer Johannes to keep > working in the CSS front, and you to make the definitive patch, but this > is for you guys to decide. I won't start on any new stuff (other than CSS) in the near future. Currently I'm trying to convert the table code to the new style handling which is a bit hairy but there are no major obstacles yet. And I really like the middle-click to submit thing. So I don't have a lot of motivation to implement real form data remembering anyway. Cheers, Johannes

Nov. 10, 2008

Greetings! -- With Patches :-)

by list-jcr＠designtools.org

On Mon, 10 Nov 2008 04:59:50 +0000 "corvid" <corvid(a)lavabit.com> wrote: > > With GTK, you can control many aspects of the toolkit such as font > > names/sizes on all applications through the system-wide or > > user-level .gtkrc files. > > I wonder whether this is why dillo has historically used > prefs.font_factor instead of a prefs.fontsize (where FontSizes in > html.cc would be generated from it). This is actually boils down to a "personal preference" issue. Some people believe it's a good thing for a web page to have fonts of multiple sizes. Being that I can be somewhat difficult, I disagree strongly, and think all text should be rendered in the exact same font size... --But of course, that's *my* "personal preference." :-) For example, header tags in HTML (i.e. H1, H2, H3, ...) typically use a larger font. If the dillo developers used prefs.fontsize, the result would be what *I* like, namely all fonts would be forced to use the exact same size. By using prefs.font_factor, the dillo developers made a wise choice because it both prevents web developers from using microscopic font sizes but still allows for variations in font size within a web page. The fonts used in the dillo user interface (menus and such), are a different beast. The older dillo (pre 2.0) used GTK which allows the user to configure user interface fonts. With the new dillo ( > 2.0), the new FLTK graphic toolkit does not allow (as far as I know) for system/user wide configuration. This means the hard coded UI fonts are stuck at the specified sizes. -- J.C. Roberts

Nov. 10, 2008

Re: patch: middle-click submits in new window/tab

by jcid＠dillo.org

On Mon, Oct 20, 2008 at 02:49:41PM +0000, corvid wrote: > I could say that it was for the sake of consistency with link behavior, > but really it was because it seemed like a dilloishly simple way > to get the equivalent of pages-remember-form-data. The consistency part is a strong argument. Also interesting is the way we can have a workaround to remember form data, before a more usual way of handling it is implemented. > > (But: who will ever discover that it exists or recognize what it > makes possible?) It may be stated in the FAQ. > 1) fltkui: it might be better to stuff all of the button event stuff > into a bag like fltkviewbase/layout do with EventButton. It certainly > looks nicer having something like a shift mask hidden away in a nice > tidy box, but I wasn't sure that it was actually worth doing. Hmm, I'd say if it's going to be a definitive patch, it's worth the effort. > 2) form: carrying around buttonNo+shifted for a long time > is not pleasing, but oh well. Agreed. Now that Johannes is interested in implementing FORM "memory", I'd say it can be stored in an URL/form-data array. I haven't checked how other browsers (e.g. FF) solve the problem of privacy (aka. when it's desirable to remember data), but most probably a quick look will shed some light. BTW, I can't compile the patch here because: make[2]: Entering directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' form.cc: In member function 'void form::Form::addButtonResource(const char*, dw::core::ui::ButtonResource*, const char*)': form.cc:241: error: cannot allocate an object of abstract type 'form::Form::FormClickedReceiver' form.hh:124: note: because the following virtual functions are pure within 'form::Form::FormClickedReceiver': ../dw/ui.hh:338: note: virtual void dw::core::ui::ButtonResource::ClickedReceiver::clicked(dw::core::ui::ButtonReso urce*, int, int) make[2]: *** [form.o] Error 1 make[2]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur/test' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/jcid/C/Dillo/d2/dillo2-cur' BTW2: * The patch changes from "Activate" to "Clicked" signal. * It also seems to change the logic in DilloHtmlInput::activate() - it makes me wonder whether pressing enter on other form widgets may submit. Anyway, it'd be good if you can coordinate with Johannes and decide what approach to take, and send a new patch for it. I'd prefer Johannes to keep working in the CSS front, and you to make the definitive patch, but this is for you guys to decide. -- Cheers Jorge.-

Nov. 10, 2008