www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - std.recls problems (dmd 0.125)

reply zwang <nehzgnaw gmail.com> writes:
When std.recls is imported, OPTLink reports:
"Warning 2: File Not Found stlp45dm_static.lib"

Besides, the documentation (std_recls.html) is totally out of date.
May 30 2005
parent reply "Matthew" <matthew hat.stlsoft.dot.org> writes:
 When std.recls is imported, OPTLink reports:
 "Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can do nothing about it.
 Besides, the documentation (std_recls.html) is totally out of date.

The documentation is, in fact, up to date. It's the code that's not. A more up-to-date version is available at http://recls.org
Jan 28 2006
next sibling parent reply S. Chancellor <dnewsgr mephit.kicks-ass.org> writes:
On 2006-01-28 18:23:02 -0800, "Matthew" <matthew hat.stlsoft.dot.org> said:

 When std.recls is imported, OPTLink reports:
 "Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can do nothing about it.
 Besides, the documentation (std_recls.html) is totally out of date.

The documentation is, in fact, up to date. It's the code that's not. A more up-to-date version is available at http://recls.org

Did he give a reason why he's refused to update it? -S.
Jan 28 2006
parent "Matthew" <matthew hat.stlsoft.dot.org> writes:
"S. Chancellor" <dnewsgr mephit.kicks-ass.org> wrote in message
news:drhgcm$qfn$1 digitaldaemon.com...
 On 2006-01-28 18:23:02 -0800, "Matthew" <matthew hat.stlsoft.dot.org>

 When std.recls is imported, OPTLink reports:
 "Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can


 nothing about it.

 Besides, the documentation (std_recls.html) is totally out of date.

The documentation is, in fact, up to date. It's the code that's not. A more up-to-date version is available at http://recls.org

Did he give a reason why he's refused to update it?

Originally it was because it was too large. I chopped some bits out for 1.5, which he didn't fancy, and so we agreed on a set of changes for version 1.6 that would make it suitable for incorporation. Version 1.6 had two rounds of "minimising" that were successful to the degree that I wrote about the experience in an instalment of my CUJ column. Alas, in the meantime I'd mooted the idea of recls2 - a total rewrite which'll allow recursively searching of any hierarchical structure - and mentioned that it might well be written in C, for maximal portability. Unfortunately, for me and for std.recls users (extant and potential), Walter decided that this was much more his cup of tea, and expressed a preference to forget our 1.6 arrangement and instead wait for recls 2. Since I've been very busy this last 12 mnths with clients and writing my next book - "Extended STL" (http://extendedstl.com/) - recls2 has not yet been written (although I have worked out all the algorithms on my paper computer). Since D is far from shippable, esp. Phobos, it seems to me that it would serve the current D community far better to have an up-to-date, if still somewhat chubby, recls 1.7.2 in Phobos - which supports FTP search, progress callbacks, and lots of other little features (e.g. ~ resolution), as well as having D-proper method casing and being in-synch with the Phobos docs - than making users work with the 27 months old version, or have the hassles of building with the latest recls as an external library. recls2 is likely to surface this year, but real life might still get in the way, so it might not. As I'm sure you can imagine, I'd rather not see users have to wait over three years to get an update. I'm not blowing smoke in saying that this is one of the reasons that my own use of D has dropped off. Every time I want to write a small file-system utility - the area where D should be an *ideal* daily tool - my decision tree offers only two realistic options, C++ or Ruby, since they have the up-to-date recls mappings. (And, though D users'll have to take my word for, recls is _incredibly_ useful. Pretty much every file-system tool I have is now (re-)implemented in terms of it.) I wouldn't mind if Phobos was an otherwise unalloyed vision of tight code with a coherent overarching paradigm. But, judging by what the NG has to say and the brief glimpses I've sneaked in recent months, it's the same stinking pile of disorganised crap (including several of my other contributions, I might add!) that it's been for the last three years. Given that, understanding Walter's position on std.recls is beyond my ken, I'm afraid. (FYI: David Friedman is going to roll the latest into the GNU variant, when he gets chance, so if you're using D on UNIX you should be free and clear in the coming weeks.) Cheers Matthew
Jan 28 2006
prev sibling parent reply =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= <afb algonet.se> writes:
Matthew wrote:

When std.recls is imported, OPTLink reports:
"Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can do nothing about it.

If it's any consolation, DMD also ships with an out of date "zlib"! Security fixed versions can be downloaded from http://www.zlib.net/ I don't see any reason to ship old versions in the standard library ? If they're not being updated anymore, then they should be left out... --anders
Jan 29 2006
parent reply J C Calvarese <technocrat7 gmail.com> writes:
In article <dri1l7$1ki6$1 digitaldaemon.com>,
=?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= says...
Matthew wrote:

When std.recls is imported, OPTLink reports:
"Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can do nothing about it.

If it's any consolation, DMD also ships with an out of date "zlib"! Security fixed versions can be downloaded from http://www.zlib.net/ I don't see any reason to ship old versions in the standard library ? If they're not being updated anymore, then they should be left out...

I agree. And it's unconscionable that the documentation for std.recls is of an incompatible version of recls. Here are the options for Walter: 1. Remove std.recls. 2. Include the new version of recls. 3. Use Ddoc to create documentation that lines up with the old version of recls. It's that simple. jcc7
Jan 29 2006
parent reply "Matthew" <matthew stlsoft.com> writes:
"J C Calvarese" <technocrat7 gmail.com> wrote in message 
news:drk5r3$1jgh$1 digitaldaemon.com...
 In article <dri1l7$1ki6$1 digitaldaemon.com>,
 =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= says...
Matthew wrote:

When std.recls is imported, OPTLink reports:
"Warning 2: File Not Found stlp45dm_static.lib"

Walter's refused to update recls for over 2 years now. I'm afraid I can do nothing about it.

If it's any consolation, DMD also ships with an out of date "zlib"! Security fixed versions can be downloaded from http://www.zlib.net/ I don't see any reason to ship old versions in the standard library ? If they're not being updated anymore, then they should be left out...

I agree. And it's unconscionable that the documentation for std.recls is of an incompatible version of recls.

I don't know that I'd go that far - these are not war crimes <g> - but it certainly doesn't help the quality of Phobos, and it's dispiriting to me, as the author of recls, to witness the unnecessary complaints and to see over 2 years' worth of additions and fixes going begging. (AFAICR, the version in phobos doesn't even do multi-part pattern matching, i.e. handling patterns of "*.d;*.cpp;*.hpp")
 Here are the options for Walter:
 1. Remove std.recls.
 2. Include the new version of recls.
 3. Use Ddoc to create documentation that lines up with the old version of 
 recls.

 It's that simple.

Obviously, I'd prefer 2. But if it's 1, I'd hope to at least have a justification as to why it's not 2. OT: I take it that no-one's taken the reins of Phobos in the last 12 months, then. Is that correct? Cheers Matthew
Jan 29 2006
next sibling parent Kyle Furlong <kylefurlong gmail.com> writes:
Matthew wrote:
 "J C Calvarese" <technocrat7 gmail.com> wrote in message 
 news:drk5r3$1jgh$1 digitaldaemon.com...
 In article <dri1l7$1ki6$1 digitaldaemon.com>,
 =?ISO-8859-1?Q?Anders_F_Bj=F6rklund?= says...
 Matthew wrote:

 When std.recls is imported, OPTLink reports:
 "Warning 2: File Not Found stlp45dm_static.lib"

do nothing about it.

Security fixed versions can be downloaded from http://www.zlib.net/ I don't see any reason to ship old versions in the standard library ? If they're not being updated anymore, then they should be left out...

of an incompatible version of recls.

I don't know that I'd go that far - these are not war crimes <g> - but it certainly doesn't help the quality of Phobos, and it's dispiriting to me, as the author of recls, to witness the unnecessary complaints and to see over 2 years' worth of additions and fixes going begging. (AFAICR, the version in phobos doesn't even do multi-part pattern matching, i.e. handling patterns of "*.d;*.cpp;*.hpp")
 Here are the options for Walter:
 1. Remove std.recls.
 2. Include the new version of recls.
 3. Use Ddoc to create documentation that lines up with the old version of 
 recls.

 It's that simple.

Obviously, I'd prefer 2. But if it's 1, I'd hope to at least have a justification as to why it's not 2. OT: I take it that no-one's taken the reins of Phobos in the last 12 months, then. Is that correct? Cheers Matthew

The best effort the community has going is the Ares project on DSource.
Jan 30 2006
prev sibling parent reply jcc7 <jcc7_member pathlink.com> writes:
In article <drkass$1qo7$1 digitaldaemon.com>, Matthew says...
"J C Calvarese" <technocrat7 gmail.com> wrote in message 
news:drk5r3$1jgh$1 digitaldaemon.com...

OT: I take it that no-one's taken the reins of Phobos in the last 12 months, 
then. Is that correct?

Walter has never let go of the reins, so how could anyone else take over? Sometimes people offer contributions. Sometimes they're partially accepted by Walter. Sometimes other people yell at them for even offering code. I think you must have been very lucky and/or persistent to get std.recls accepted in the first place. The Ares project is our best hope, but it isn't ready yet to take the place of Phobos yet. jcc7
Jan 30 2006
parent "Matthew" <matthew hat.stlsoft.dot.org> writes:
"jcc7" <jcc7_member pathlink.com> wrote in message
news:drlg8l$aiq$1 digitaldaemon.com...
 In article <drkass$1qo7$1 digitaldaemon.com>, Matthew says...
"J C Calvarese" <technocrat7 gmail.com> wrote in message
news:drk5r3$1jgh$1 digitaldaemon.com...

OT: I take it that no-one's taken the reins of Phobos in the last 12


then. Is that correct?

Walter has never let go of the reins, so how could anyone else take over?

But is this because he's waiting for the right moment/recipient, or he never intends to? If it's the latter, then it's going to make it hard for D to be anything other than a compiler-writer's testbed. If it's the former, is there any hint of a plan as to when/whom?
 Sometimes people offer contributions. Sometimes they're partially accepted

 Walter. Sometimes other people yell at them for even offering code. I

 must have been very lucky and/or persistent to get std.recls accepted in

 first place.

I guess. But in retrospect it's had more negative side-effects than positive ones. People get the mistaken impression that recls is broken, or not much cop, and I personally have given up on trying to use D for anything real, because even though there _is_ a way to make the real recls/D work with phobos, it's abstruse to the degree that I forget it every time, and am confused the next time I try. Compared to the total doddle that is writing such apps in Ruby, using D is just a nonsensical alternative.
 The Ares project is our best hope, but it isn't ready yet to take the

 Phobos yet.

Yes. Sean's been helping me with some of my stuff recently, and I hope to be able to pay him back with some help on Ares later this year. Maybe std.recls can go into Ares, and I can just forget Phobos as a bad dream.
Jan 31 2006