[typeof]extremely long lines in vector200.hpp causes ClearCase problem
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase which has an 8000-character upper limit for text files. The workaround is to check it in as a binary file but this eliminates the possibility of future merges. Is it possible to make the lines shorter? Thanks, Sean
Sean Huang wrote:
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase which has an 8000-character upper limit for text files. The workaround is to check it in as a binary file but this eliminates the possibility of future merges. Is it possible to make the lines shorter?
Not easily - it's preprocessor generated code. Running some kind of code beautifier on it might work (meant as a user-side solution). I once mailed a replacement for the current header generator to Peder and Arkadiy that uses the Wave tool (but there were some concurrent code changes and performance issues with Wave at that time, so it never made it to the CVS) and Hartmut mentioned that he would maybe add a pretty-printing option to the Wave tool, some day. That's one possibility for a Boost-side solution. Another one would be to rewrite the code generator to use more vertical repetition. Regards, Tobias
----- Original Message ----- From: "Tobias Schwinger" <tschwinger@neoscientists.org>
Sean Huang wrote:
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase which has an 8000-character upper limit for text files. The workaround is to check it in as a binary file but this eliminates the possibility of future merges. Is it possible to make the lines shorter?
Not easily - it's preprocessor generated code. Running some kind of code beautifier on it might work (meant as a user-side solution). Thanks for the explanation.
I am curious why the author(s) did not leave the preprocessor directives in place. Not willing to introduce dependencies on boost.preprocessor? The file is very well formatted so I suspect it has been beautified. Maybe it was just an oversight. Sean
Sean Huang wrote:
----- Original Message ----- From: "Tobias Schwinger" <tschwinger@neoscientists.org>
Sean Huang wrote:
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase which has an 8000-character upper limit for text files. The workaround is to check it in as a binary file but this eliminates the possibility of future merges. Is it possible to make the lines shorter?
Not easily - it's preprocessor generated code. Running some kind of code beautifier on it might work (meant as a user-side solution).
Thanks for the explanation.
I am curious why the author(s) did not leave the preprocessor directives in place. Not willing to introduce dependencies on boost.preprocessor?
It was me (not being an author of that library) who contributed that part (that's why I replied). It's an optimization that speeds up preprocessing by about factor 10. Regards, Tobias
Sean Huang wrote:
----- Original Message ----- From: "Tobias Schwinger" <tschwinger@neoscientists.org>
Sean Huang wrote:
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase which has an 8000-character upper limit for text files. The workaround is to check it in as a binary file but this eliminates the possibility of future merges. Is it possible to make the lines shorter?
That file is read only if BOOST_TYPEOF_LIMIT_SIZE is set to >= 200. That macro currently defaults to 50. So your copy of Boost will still work if you omit that file. Regards, Tobias
---- Original Message ----- From: "Tobias Schwinger" <tschwinger@neoscientists.org>
That file is read only if BOOST_TYPEOF_LIMIT_SIZE is set to >= 200. That macro currently defaults to 50. So your copy of Boost will still work if you omit that file.
Thank you very much for the explanation. Again, would you consider making the lines shorter since it should not causes any problem and you probably already run a beautifier on it. We can fix it every time we take a boost CVS snapshot and import it into our source control system but it would be very tedious. Best regards, Sean
"Sean Huang" <huangsean@hotmail.com> wrote
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase
Just out of curiosity -- why would one want to put this file into ClearCase? Regards, Arkadiy
----- Original Message ----- From: "Arkadiy Vertleyb" vertleyb@hotmail.com
"Sean Huang" <huangsean@hotmail.com> wrote
vector200.hpp contains lines with more than 8000 characters. This causes problems when the file is added to IBM Rational ClearCase
Just out of curiosity -- why would one want to put this file into ClearCase?
The same reason the file is in CVS. We do not use Boost as a shrink-wrapped product and need to customize, patch, and delivery Boost to our customers inside our organization. Boost release cycles do not really work that well for us so we do internal testing and releases. Maybe I misunderstood your question. Sean
"Sean Huang" <huangsean@hotmail.com> wrote
Just out of curiosity -- why would one want to put this file into ClearCase?
The same reason the file is in CVS. We do not use Boost as a shrink-wrapped product and need to customize, patch, and delivery Boost to our customers inside our organization. Boost release cycles do not really work that well for us so we do internal testing and releases.
I see. I don't think we can do anything before 1.34 is out. After this we can look at this. Are you having the same issue with vector150.hpp? How complex are you types --can you get by with max LIMIT_VECTOR_SIZE of 150? Regards, Arkadiy
----- Original Message ----- From: "Arkadiy Vertleyb" <vertleyb@hotmail.com> To: <boost@lists.boost.org> Sent: Wednesday, December 20, 2006 12:06 PM Subject: Re: [boost] [typeof]extremely long linesinvector200.hppcausesClearCase problem
"Sean Huang" <huangsean@hotmail.com> wrote
I see.
I don't think we can do anything before 1.34 is out. After this we can look at this.
Thanks!
Are you having the same issue with vector150.hpp? How complex are you types --can you get by with max LIMIT_VECTOR_SIZE of 150?
Our import script only complained about vector200.hpp since the upper limit is 8000 characters. 150 would be fine for us. Still I'm a bit puzzled by why it is so hard to shorten the lines. Best regards, Sean
"Sean Huang" <huangsean@hotmail.com> wrote
Our import script only complained about vector200.hpp since the upper limit is 8000 characters. 150 would be fine for us. Still I'm a bit puzzled by why it is so hard to shorten the lines.
I didn't say it's *very* hard, but this is the preprocessor-generated code -- we would have to somehow modify the vector template generation -- as Tobias said -- maybe use verticle repitition, etc... Or maybe modify the perl script Tobias wrote... But, we are in the code freeze right now, so I would prefer to avoid *any* changes until 1.34 is out. So, if just not storing vector200.hpp, and limiting your vector size to 150 works for you, let's wait until 1.34 is out. Regards, Arkadiy
----- Original Message ----- From: "Arkadiy Vertleyb" <vertleyb@hotmail.com>
"Sean Huang" <huangsean@hotmail.com> wrote
But, we are in the code freeze right now, so I would prefer to avoid *any* changes until 1.34 is out. So, if just not storing vector200.hpp, and limiting your vector size to 150 works for you, let's wait until 1.34 is out.
That sounds reasonable. Thank you very much for the help! Sean
participants (3)
-
Arkadiy Vertleyb -
Sean Huang -
Tobias Schwinger