- 28 Apr, 2016 2 commits
-
-
POTTIER Francois authored
-
POTTIER Francois authored
-
- 05 Nov, 2015 1 commit
-
-
POTTIER Francois authored
-
- 04 Nov, 2015 1 commit
-
-
POTTIER Francois authored
Updated the table back-end to properly initialize the position in the sentinel cell at the bottom of the stack. This requires changing the function [start] in the incremental API to take a position. Updated the doc and demos.
-
- 02 Nov, 2015 1 commit
-
-
POTTIER Francois authored
-
- 29 Oct, 2015 1 commit
-
-
POTTIER Francois authored
For the modules in MenhirLib, adopt the convention that the first letter of the file name must be capitalized.
-
- 19 Oct, 2015 3 commits
-
-
POTTIER Francois authored
-
POTTIER Francois authored
-
POTTIER Francois authored
-
- 16 Oct, 2015 4 commits
-
-
POTTIER Francois authored
-
POTTIER Francois authored
-
POTTIER Francois authored
-
POTTIER Francois authored
-
- 29 Sep, 2015 1 commit
-
-
POTTIER Francois authored
-
- 06 Jul, 2015 1 commit
-
-
POTTIER Francois authored
-
- 22 Jan, 2015 1 commit
-
-
POTTIER Francois authored
Modify [Engine] to give access to the positions of the lookahead token. Use a tainted dummy position to help debug.
-
- 18 Jan, 2015 2 commits
-
-
POTTIER Francois authored
-
POTTIER Francois authored
-
- 14 Jan, 2015 1 commit
-
-
POTTIER Francois authored
Defined the type [stack] as an abbreviation for [element stream].
-
- 13 Jan, 2015 1 commit
-
-
POTTIER Francois authored
(Not great, but they don't deserve a separate module yet.) Updated Printers and Calc.
-
- 10 Jan, 2015 1 commit
-
-
POTTIER Francois authored
This removes the need for the exception [Accept]. Modified the table back-end to include the number of start productions as part of the generated tables.
-
- 29 Dec, 2014 2 commits
-
-
POTTIER Francois authored
The type env no longer needs to be parameterized. This makes the API safe (and simpler). There a slight runtime cost (e.g., from 4.6s to 4.7 seconds).
-
POTTIER Francois authored
This seems to cause roughly a 10% speed loss (e.g. from 4.3 to 4.6 seconds). The current interface is not satisfactory, as it allows the user to manufacture a continuation that will break our invariants and lead to a crash.
-
- 27 Dec, 2014 1 commit
-
-
POTTIER Francois authored
[lr1state], [element], [stream] and of the function [view].
-
- 24 Dec, 2014 1 commit
-
-
POTTIER Francois authored
-
- 11 Dec, 2014 2 commits
-
-
POTTIER Francois authored
in the special case where the language is empty or {epsilon} is now fixed.
-
POTTIER Francois authored
This removes two eta-redexes. The speed difference is barely measurable, maybe 0-5%.
-
- 09 Dec, 2014 2 commits
-
-
POTTIER Francois authored
-
POTTIER Francois authored
This costs one more field in the [env] record. The speed difference is actually not measureable. Maybe it would be measureable in an application where the [env] records are stored and move to the major heap. Never mind.
-
- 08 Dec, 2014 7 commits
-
-
POTTIER Francois authored
to artifically cast [offer] and [handle] to a more general type.
-
POTTIER Francois authored
Added a phantom type parameter to the type [env] so as to prevent the user from confusing [offer] and [handle], thus eliminating a runtime check, while keeping the ability of offering functions that operate on an arbitrary environment (such as stack inspection).
-
POTTIER Francois authored
every step of error handling. Compared to Frédéric's patch, this code is "more incremental" in that the parser will try to handle the error, pop one cell, suspend itself, try to handle the error, pop one cell, suspend itself, etc. In other words, *every call* to [error env] has been replaced with [HandlingError env], whereas in Frédéric's patch only 2 out of 3 calls were replaced.
-
POTTIER Francois authored
-
POTTIER Francois authored
-
POTTIER Francois authored
-
POTTIER Francois authored
For convenience, this is done using [match/exception]. As a result, ocaml 4.02 is required. Updated the Makefile to reflect this requirement.
-
- 06 Dec, 2014 3 commits
-
-
POTTIER Francois authored
The first token is now read during the first call to [run]. This exchanges the order of the first two messages in the trace, but is actually more regular. In the code back-end, this fixes a "bug" in the case where the grammar accepts only the empty word. In the table back-end, the bug is not yet fixed.
-
POTTIER Francois authored
Replaced the call to [read] with returning an [InputNeeded] result.
-
POTTIER Francois authored
I see a 5% performance gain, from 2.0 to 1.9. If we could remove the tests altogether, we would go down to 1.8.
-
- 05 Dec, 2014 1 commit
-
-
POTTIER Francois authored
The ratio table/code seems to be back from 2.1 to 2.0.
-