[TxMt] Bookmarks

[Ralph Pöllath]

On 28.01.2005, at 06:31, Charles Miller wrote:
> On 27/01/2005, at 10:19 PM, Ralph Pöllath wrote:
>> IMO, bookmarks cannot be implemented properly (go ahead Allan, prove 
>> me wrong :-). Automatic bookmarks (that "boomark" each line matching 
>> a certain pattern), would be much more useful.
>
> It's a common misconception that if there are significant situations 
> in which some idea won't work, it's "impossible". Often, though, 
> having something work most of the time is still better than not having 
> it at all.
>
> * Store bookmarks in the project as line number, and some of the 
> context surrounding the bookmark.
> * If, on loading the file, the number and context don't match, start 
> looking up and down for the first line that matches the saved context 
> to a certain degree of tolerance.
> * If some bookmarks are still unaccounted for, delete them and pop up 
> a dialog saying "This file was modified since it was last opened in 
> [[Project Name]]. Some of your bookmarks have been removed."

Yup, that's what I referred to as "merging".

> Under certain circumstances this would lead to bookmarks being lost 
> (or even moved), but it would work well enough for 90% of cases.

I guess it boils down to a matter of taste. I don't deny that all of 
this would be helpful for some or even most people, but personally, I 
wouldn't use e.g. a filesystem that let's me get at my data only 90% 
percent of the time. Sometimes, and for some people, it's those 10% 
that count.

Additionally, you and I know why that system wouldn't work in some 
small percentage of cases. But what about those who don't know or care 
about the implementation?

I could be wrong, but I assume that TM is mainly used by power users, 
and I further assume that those tend to work with structured text (like 
code, xml, etc) - that's why we have all these bundles. So IMO, it 
would make a lot more sense to exploit the text's inherent structure 
for navigation. It works 100% percent of the time even for non-project 
files, doesn't require storing extra data, and rewards you for 
structuring your text :-)

Cheers,
-Ralph.