Re: Some hints for usability improvements

On Thu, 19 Jan 2006 18:50:50 -0500
Peter Bloomfield <peterbloomfield bellsouth net> wrote:

> Hi Mişu!
> On 01/19/2006 04:23:03 PM, Mişu Moldovan wrote:
> > 
> >  First of all forgive me for the blasphemy of using Sylpheed to post   
> > to  Balsa's mailing list.
> Oh, this is a very inclusive, forgiving, list!
> > I'm using an old computer with very limited memory right now and I  
> > prefer to use Sylpheed here because of the  lighter memory footprint.
> Do you have statistics on the footprint?  We've tried to make Balsa  
> lighter...

I'll follow up on this on another thread.

> > Also, the latest Balsa compiled on this  oldie, but crashes on  
> > startup (I'll investigate this some other day).
> We had some reports of problems with old glib/gtk installations--what  
> versions were involved?

2.6.4/2.6.4 as included in Debian stable (aka Sarge). Balsa is 2.3.8. Ok, I have just started it again and it doesn't crash anymore. It crashes only when previewing a certain mail which happened to be the last one in my Inbox at that time (and having the preference to automatically open Inbox and automatically open a mail, it crashed Balsa on startup).  I've also noticed that this mail doesn't crashes the same version of Balsa on a Gentoo system which has a 2.8.8 gtk and a 2.8.4 glib. The mail in mbox format for the curious may be downloaded from: (just to be sure)
> > Although I still prefer Balsa 2.3 over Sylpheed 2.2 or Sylpeed Claws
> > 2.0, here are some little things from Sylpheed that would (in my
> > opinion) improve the usability of Balsa:
> > 
> >  1) Balsa's message editor could use an undo history. For now there's  
> > a  single level of undo and it works oddly, the cursor is moved out  
> > of  place when doing an undo. Contrast this with the configurable  
> > number  of undo steps in Sylpheed. (Ok, ok, maybe we don't need a  
> > preference  for this, but 50 levels of undo/redo should be enough for  
> > anyone,  especially if everything works right).
> With current cvs, if you have GtkSourceView and configure  
> --with-gtksourceview, you get multilevel undo (not configurable, it's  
> GtkSourceView's default setting, which Iirc is 25).  You also get  
> highlighting of /structured phrases/, and quoted line coloring in the  
> compose window.

Sound fine, I'll give it a try once the current cvs is released as a new version.

> >  2) Replying to a mail with several levels of replies breaks the   
> > paragrafs lines in crude ways. Granted, you may reflow manually by   
> > selecting the affected paragraphs and reflowing them manually with   
> > Control-R or the corresponding menu entry. But the logic of doing it   
> > automatically is not so complex. Sylpheed seems to get it right (and   
> > I  think it's not the only MUA to do so).
> Balsa handles it when the message you're answering has format=flowed,  
> which afaict is the only format that allows a MUA to know when text is  
> in paragraphs that should be reflowed, as opposed to arbitrary text  
> (code snippets, etc.) that shouldn't.

For this and other format=flowed related problems I'll open up another thread later.

> >  3) Balsa's spell checker is primitive. I rarely use it because it's   
> > hard for me to follow it as it checks through the text. There seems   
> > to  be no indication to what word is being questioned and you have to  
> > guess a lot. (Disclaimer: This message is not spell checked, but at   
> > work I've tried Sylpheed Claws with spell checking and it evens   
> > checks  the spelling as you write).
> Since 2.3.7, if you have GtkSpell and configure --with-gtkspell, you  
> get, well, real spell-checking.  The spell-check option (toolbar button  
> and menu item) becomes a toggle, and spell-checking is as-you-type.   
> We're still working on saving the state of spell-checking when a  
> message is saved to and continued from Drafts.

Tried that and it works quite nice. Why isn't it enabled by default? It seems to be much better than the old spell checker.

> >  4) Balsa, please be nice and ask me before sending a mail with an   
> > empty Subject header...
> Yes, that would be kind!
> > And also ask me about automatically  re-encoding to UTF-8 if my mail  
> > contains characters not present in  the  default encoding.
> The current code tries quite hard to find a charset that can handle the  
> text, but yes, it could ask for suggestions before falling back on  
> UTF-8.

Actually, I was thinking about just asking for a confirmation for falling to UTF-8 when the supposed encoding (set as default or set in the mail to which Balsa composed the reply) doesn't fit.

> >  5) As someone that usually reads only the new mails on several   
> > folders  and subfolders in an IMAP hierarchy I frequently navigate  
> > from one  new  message to another using the Ctrl-N shortcut. When  
> > there are no new  messages in the current folder I have to manually  
> > switch to a folder  with some new mails to continue reading them.  
> > Sylpheed does what I  mean and asks me if it should change  
> > automatically to the next new  mail in the next subfolder in my IMAP  
> > hierarchy. True also for other  MUA's.
> Maybe pop up a dialog saying "Next unread message is in %s" with "OK"  
> and "Cancel" buttons?

That would rock.

> >  6) In both Sylpheed MUA's it's possible to set up the folder pane to  
> > only show the folders and the number of new mails in parantheses   
> > along  side the folder's name. No need to display the whole list of   
> > statistics just to see how many new mails you have in a specific IMAP  
> > sub-folder. As a side note, I got to distrust the new status of IMAP   
> > subfolders in Balsa especially with some (maybe broken) IMAP servers   
> > like Exchange.
> Yes, that would be a useful option.
> > 7) Sylpheed Claws manages to briliantly organize the multitude of
> > options from Sylpheed preferences in a window with chapters and
> > subchapters on the left side (no more sub-tabs in a tab of the
> > Preferences window).
> Is that a custom widget?  I don't know of a Gtk widget that would do  
> that...

I'm posting a screenshot at
and here's a (hopefully useful) snippet from Sylpheed-Claws's source code:

struct _PrefsWindow
        GtkWidget *window;
        GtkWidget *table1;
        GtkWidget *scrolledwindow1;
        GtkWidget *scrolledwindow2;
        GtkWidget *tree_view;
        GtkWidget *table2;
        GtkWidget *pagelabel;
        GtkWidget *labelframe;
        GtkWidget *frame;
        GtkWidget *notebook;
        GtkWidget *confirm_area;
        GtkWidget *ok_btn;
        GtkWidget *cancel_btn;
        GtkWidget *apply_btn;
        gint *save_width;
        gint *save_height;

        GtkWidget *empty_page;

        gpointer         data;
        GSList          *prefs_pages;
        GtkDestroyNotify func;

struct _PrefsTreeNode
        PrefsPage *page;
        gfloat     treeweight; /* GTK2: not used */



Attachment: pgpyc3h7Guqzm.pgp
Description: PGP signature

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]