> Here's what I can think of at the moment:
> * There's inconsistent use of int vs. ws_int_t (and others I'd
> guess). I've seen it mostly in WinSTL, though its probably
> elsewhere also.
> * There's inconsistent use of ::SendMessage vs winstl_send_message
> (or something likely, don't remember exactly right now)

Not sure either of these is a huge problem, but I agree that unmotivated inconsistency is a motivation to make consistent. ;)

> * I think the new C-style headers have a slight (techincal) problem with the standard, in that they use double underscores as substitute for the scope operator, but then double underscore names are reserved by the standard AFAIK

I was about to reply "that's only for leading underscores", and verified such in the C99 standard. Then I looked in C++98, and was stunned to learn that the implementation reserved indentifiers containing _embedded_ double underscores. All I can say is: what a ridiculous rule!

Since the C-compatible functions are already out there, I'm not going to change them.

> * I believe some [winstl, don't know about other] sequence classes could be refactored to remove lots of duplicate nearly identical code, this is how I've tried to write the sequences with string/value pairs (e-mail coming in a few hours)

I don't doubt it. If you can find specific examples where you suspect this may be so, I can get to it. This will also help me with "Extended STL" as I'll be covering the chapters on writing collections over the next 2-3 weeks.

> * The th_*_sequence classes are the only contribution I'd say is missing
>
>> (And some help on the docs won't hurt <g>)
>
> You say you've been upgrading some documentation already... which is it so I can take a look?

It's the 1.9 stuff. I'll email you a chm in a few days when it's a bit more in shape.

> > * There's inconsistent use of ::SendMessage vs winstl_send_message (or something likely, don't remember exactly right now)
>
> Not sure either of these is a huge problem, but I agree that unmotivated inconsistency is a motivation to make consistent. ;)

Right... Well, you asked for inconsistencies...

> I was about to reply "that's only for leading underscores", and verified such in the C99 standard. Then I looked in C++98, and was stunned to learn that the implementation reserved indentifiers containing _embedded_ double underscores. All I can say is: what a ridiculous rule!

While I haven't even begun to try to read the standard, the underscore rules are "Sutter 101" for me (amongst all the things I could've learned from reading his books, that's one of few that really stuck with me).

> > * I believe some [winstl, don't know about other] sequence classes could be refactored to remove lots of duplicate nearly identical code, this is how I've tried to write the sequences with string/value pairs (e-mail coming in a few hours)
>
> I don't doubt it. If you can find specific examples where you suspect this may be so, I can get to it. This will also help me with "Extended STL" as I'll be covering the chapters on writing collections over the next 2-3 weeks.

Will do...


Pablo

"Pablo Aguilar" <paguilarg@hotmail.com> wrote in message news:dclvku$chp$1@digitaldaemon.com...
> Ok, we moved at work today to the new release...
>
> Here's the 2 things I've found so far:
>
> 1) listbox_sequence (and I'd say all of it's kin) is still missing
> reference/const_reference typedefs, that are required by sequence_range.

Fixed. Included. (The sequences are still non-mutating ones - I'll be doing mutating versions this/next week, as part of my "Extended STL" research - so they only contain const_reference. That should be enough to satisfy sequence_range, however.)

> 2) c_str_ptr(*findfile_sequence::iterator) now returns a file's name, and
> not it's path (it used to return it's full path). Now, there's likely a very
> good reason for this, but for the time being it broke some code

Already dealt with.

:-)